Artificial intelligence-driven discovery of coumarin-based therapeutics: Revolutionizing target identification and validation.

Artificial intelligence-driven discovery of coumarin-based therapeutics: Revolutionizing target identification and validation - PMC Skip to main content An official website of the United States government Here's how you know Here's how you know Official websites use .gov A .gov website belongs to an official government organization in the United States. Secure .gov websites use HTTPS A lock ( Lock Locked padlock icon ) or https:// means you've safely connected to the .gov website. Share sensitive information only on official, secure websites. Search Log in Dashboard Publications Account settings Log out Search… Search NCBI Primary site navigation Search Logged in as: Dashboard Publications Account settings Log in Search PMC Full-Text Archive Search in PMC Journal List User Guide PERMALINK Copy As a library, NLM provides access to scientific literature. Inclusion in an NLM database does not imply endorsement of, or agreement with, the contents by NLM or the National Institutes of Health. Learn more: PMC Disclaimer | PMC Copyright Notice Pharm Sci Adv . 2026 Apr 9;4:100119. doi: 10.1016/j.pscia.2026.100119 Search in PMC Search in PubMed View in NLM Catalog Add to search Artificial intelligence-driven discovery of coumarin-based therapeutics: Revolutionizing target identification and validation Yasser Fakri Mustafa Yasser Fakri Mustafa 1 Department of Pharmaceutical Chemistry, College of Pharmacy, University of Mosul, Mosul, 41001, Iraq Find articles by Yasser Fakri Mustafa 1 Author information Article notes Copyright and License information 1 Department of Pharmaceutical Chemistry, College of Pharmacy, University of Mosul, Mosul, 41001, Iraq Received 2026 Jan 27; Revised 2026 Apr 7; Accepted 2026 Apr 7; Collection date 2026 Dec. © 2026 The Author This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/). PMC Copyright notice PMCID: PMC13092200  PMID: 42011244 Abstract Artificial intelligence (AI) has transformed modern drug discovery by reshaping how therapeutic targets are identified, validated, and optimized. Coumarin derivatives, with their diverse pharmacological activities and structural adaptability, offer a rich chemical space for AI-guided exploration. However, despite the potential of integrating AI methodologies with coumarin chemistry to accelerate the identification of novel disease targets and design safer, more efficient drug candidates, a unified synthesis of these data-driven workflows remains lacking. This review aims to address this research gap by compiling and analyzing literature from the past ten years to provide a comprehensive content framework. The objective is to evaluate data-driven approaches—ranging from literature-based data mining, molecular docking, and predictive modeling to deep-learning frameworks and multiomics integration—that collectively enhance coumarin-target discovery. Emphasis is placed on AI-enabled workflows that connect structural, functional, and phenotypic data to support hypothesis generation, target prioritization, and validation across computational and experimental domains. Recent studies demonstrate that AI-assisted algorithms can accurately predict coumarin–protein interactions, uncover unrecognized biological targets, and rationalize structure–activity relationships. Deep-learning and risk–benefit models have improved target ranking, while multiomics data fusion has revealed disease-specific mechanisms in oncology, metabolic, infectious, and cardiovascular disorders. These insights have translated into tangible outcomes, such as the design of novel coumarin–quinone hybrids and selective enzyme inhibitors. The convergence of AI and coumarin-based medicinal chemistry heralds a paradigm shift in therapeutic target identification. Future research directions and prospects should focus on ethical data governance, interpretability, and cross-disciplinary collaboration to position AI-driven coumarin research at the forefront of next-generation precision therapeutics. Keywords: Artificial intelligence, Coumarins, Therapeutic targets, Machine learning, Molecular docking, Multiomics Graphical abstract Open in a new tab Highlights • AI accelerates the discovery of therapeutic targets for diverse coumarin derivatives. • Integrated data ecosystems enhance coumarin target prediction and prioritization. • Knowledge graphs reveal coumarin–target networks across biological scales. • Deep learning refines coumarin target hypotheses with greater translational value. • AI-driven workflows reduce risk and shorten timelines in coumarin drug discovery. List of abbreviations ADMET Absorption, Distribution, Metabolism, Excretion, and Toxicity AI Artificial Intelligence CNS Central Nervous System CRISPR–Cas9 Clustered Regularly Interspaced Short Palindromic Repeats–Associated Protein 9 DNA Deoxyribonucleic Acid EMA European Medicines Agency FDA Food and Drug Administration GAN Generative Adversarial Network GNN Graph Neural Network LD/AGC Ligand-Directed/Affinity Generation Chemistry ML: Machine Learning MSE Mean Squared Error QSAR Quantitative Structure–Activity Relationship R 2 Coefficient of Determination RMSE Root Mean Squared Error RNA Ribonucleic Acid ROC–AUC Receiver Operating Characteristic – Area Under the Curve SAR Structure–Activity Relationship SARS-CoV-2 Severe Acute Respiratory Syndrome Coronavirus 2 SVM Support Vector Machine VAE Variational Autoencoder XAI Explainable Artificial Intelligence 1. Introduction The rapid development of effective therapeutic interventions relies on the precise identification and thorough evaluation of novel biological targets, including those beyond the traditional validated domains [ 1 ]. Recent progress in artificial intelligence (AI) has revolutionized this process, particularly through advancements in algorithm design and data-driven modeling. AI now enables automated analysis of complex patient-derived datasets, facilitating the identification, prioritization, and validation of potential therapeutic targets. Through predictive modeling, knowledge graphs, and generative algorithms, AI enhances human creativity by guiding scientific inquiry toward the most promising and impactful directions [ 2 ]. In medicinal chemistry, coupling AI-driven analytics with the rich chemical diversity of coumarin derivatives has opened new frontiers for drug discovery. The exploration of structure–activity relationships within coumarin scaffolds has yielded numerous bioactive leads with applications across diverse therapeutic fields [ 3 ]. Modern tools such as automated molecular docking, deep learning, data mining, and multiomics integration—the combined analysis of multiple biological data layers such as the genome, proteome, and metabolome—now allow AI to bridge computational predictions with experimental validation. These integrative approaches can help streamline early-stage discovery pipelines and may reduce certain developmental risks by enabling preliminary predictions of efficacy- and safety-related properties during the initial phases of drug development [ 4 ]. The synergy between AI and coumarin chemistry represents a transformative platform for small-molecule drug discovery. The coumarin nucleus serves as an adaptable scaffold for rational design [ 5 ]. When coupled with AI's capacity to process vast chemical and biological datasets, this scaffold enables rapid hypothesis generation, virtual screening, and optimized synthesis routes. With robust data curation, risk assessment, and model refinement, AI-driven strategies are redefining how coumarin-based compounds are designed, evaluated, and advanced across major disease domains [ 6 ]. Beyond accelerating established discovery workflows, an important conceptual question concerns whether AI fundamentally changes the nature of therapeutic discovery or primarily improves its efficiency. Historically, coumarin-based drug discovery relied on empiricism-driven medicinal chemistry, where structural modification, biological screening, and incremental optimization were guided largely by chemical intuition and prior pharmacological knowledge. In contrast, AI-enabled discovery introduces a data-centric paradigm in which large chemical, biological, and clinical datasets are simultaneously interrogated to reveal non-obvious relationships between molecular scaffolds and biological systems [ 7 ]. Rather than functioning solely as an accelerative computational infrastructure, certain AI frameworks operate as hypothesis-generating engines capable of identifying previously unrecognized coumarin–target associations, predicting multi-target activity profiles, or proposing scaffold modifications that extend beyond traditional medicinal chemistry heuristics. For example, machine-learning-guided scaffold exploration can identify coumarin analogues with unexpected binding affinity toward unrelated protein families, while network-based algorithms may reveal polypharmacological target constellations relevant to complex diseases such as cancer or neurodegeneration [ 8 ]. These capabilities suggest that AI does not merely accelerate discovery timelines but may also expand the conceptual search space of coumarin pharmacology by uncovering mechanistic hypotheses that would be difficult to derive through classical experimental reasoning alone [ 9 ]. Despite these advances, the role of AI in drug discovery remains an evolving paradigm rather than a universally validated solution. While AI has demonstrated considerable potential in accelerating data analysis and hypothesis generation, its integration into medicinal chemistry workflows continues to be critically evaluated. Questions remain regarding model interpretability, data quality, and the extent to which computational predictions translate into clinically meaningful outcomes. Consequently, AI have to be viewed not merely as an inevitable technological progression but as a methodological framework whose strengths and limitations must be carefully examined within the context of experimental pharmacology and chemical biology [ 10 ]. The literature included in this review was identified through searches of major scientific databases including PubMed, Scopus, and Web of Science. Keywords such as “coumarin”, “artificial intelligence”, “machine learning”, “deep learning”, “molecular docking”, “network pharmacology”, and “multiomics” were used individually and in combination to retrieve relevant publications. Priority was given to peer-reviewed articles reporting AI-assisted drug discovery methodologies, computational modeling approaches, and pharmacological studies involving coumarin derivatives. Additional references were identified through citation tracking of relevant reviews and primary research articles. The final selection focused on studies that illustrate the integration of AI techniques with medicinal chemistry strategies for target identification, molecular optimization, and therapeutic evaluation of coumarin-based compounds. 2. Medicinal chemistry and historical perspective of coumarins Coumarins are a class of oxygen-containing heterocyclic compounds with a benzopyrone core. They have been of great interest in medicinal chemistry because they are easy to synthesize, have a simple structure, and have a wide range of biological activity [ 11 ]. The use of coumarins in medicine goes back to the early 1800s, when Dipteryx odorata (tonka beans) was first used to make coumarin. Coumarins were first known for their odor, but they quickly became important in medicine when it was found that they could hinder blood from coagulating. The discovery led to the creation of warfarin, one of the first and most successful small-molecule medications made from this scaffold [ 12 ]. Coumarin is extremely valuable for chemical modification from a medicinal chemistry point of view. Chemical substitutions at the 3-, 4-, 6-, 7-, and 8-positions allow for precise changes to physicochemical characteristics, target affinity, and pharmacokinetic responses [ 13 ]. The chemical structure of the coumarin nucleus is shown in Fig. 1 . Over the years, systematic synthetic work has led to the discovery of many natural, semi-synthetic, and completely synthetic coumarin derivatives that have anticoagulant [ 14 ], antibacterial [ 15 ], anticancer [ 16 ], anti-inflammatory [ 17 ], neuroprotective [ 18 ], antiviral [ 19 ], and enzyme-inhibitory effects [ 20 ]. In light of this variety, coumarins are a suitable place to commence lead finding and optimization programs. Fig. 1. Open in a new tab Chemical structure of coumarin nucleus. In the past, finding new drugs based on coumarin relied mainly on biological screening and changing the structure of the chemical based on what was known. Improvements in synthetic organic chemistry made it possible to make fused coumarins, heterocycle-linked hybrids, and substituted analogues [ 21 , 22 ]. This gave medicinal chemists a logical way to study structure–activity relationships (SAR). Despite significant synthetic research, several potential coumarin derivatives have not advanced beyond preclinical stages due to challenges like off-target effects, metabolic instability, low solubility, or toxicity [ 23 ]. In recent years, it has been easier to get information on biology, chemistry, and pharmacology. This has caused coumarin research to move toward more in-depth, data-driven methodologies. This shift makes it possible for computational and AI-based technologies to get involved [ 24 ]. These approaches can help classical medicinal chemistry by helping to improve scaffolds, choose which replacements to make first, and uncover new therapeutic targets. These new methodologies don't replace chemistry-driven design; instead, they build on the huge quantity of information that has been collected over time on the coumarin scaffold [ 25 ]. Despite the growing number of studies reporting diverse biological activities of coumarin derivatives, a critical examination of the literature reveals several limitations that may influence the interpretation of these findings. Many studies rely on relatively small compound libraries or focus on specific substitution patterns of the coumarin scaffold, which may restrict the chemical diversity explored in computational models. In addition, variations in experimental assay conditions and biological models can lead to inconsistencies in reported activity profiles, complicating direct comparison across studies. From a computational perspective, AI-based predictive models trained on such heterogeneous datasets may capture correlations that are not necessarily mechanistically meaningful [ 26 ]. Consequently, while coumarins remain attractive scaffolds for AI-guided drug discovery due to their structural versatility and broad pharmacological potential, careful data curation, standardized experimental validation, and integration of mechanistic insights remain essential for translating computational predictions into robust therapeutic leads. 3. Molecular mechanisms of action and Structure–Activity relationships of Coumarin derivatives Coumarin derivatives demonstrate their pharmacological potential by interacting with diverse biological targets. It has been shown that coumarins inhibit vital proteins at the enzyme level, among them carbonic anhydrases [ 27 ], monoamine oxidases [ 28 ], topoisomerases [ 29 ], kinases [ 30 ], and acetylcholinesterase [ 31 ]. The coumarin core and active-site residues most often interact through π–π stacking, hydrogen bonding, and hydrophobic interactions [ 32 ]. Coumarin compounds demonstrate anticancer potential through numerous mechanisms, including the inhibition of cell cycle-regulating enzymes, alteration of apoptotic pathways, distortion of DNA topology, and suppression of angiogenesis [ 33 ]. The patterns of substitution on the coumarin nucleus are particularly important for the specificity and selectivity of molecular targets. As an instance, electron-donating groups at the 7-position are commonly connected to higher antiproliferative action. On the other hand, bulky substituents or fused heterocycles can make binding affinity stronger for specific cancer-related enzymes [ 34 ]. Coumarins show their antibacterial and antiviral effects by inhibiting enzymes, breaking down membranes, and interfering with the formation of nucleic acids [ 35 ]. Similarly, anti-inflammatory effects are often linked to the regulation of oxidative stress pathways, the inhibition of cyclooxygenases, or the attenuation of pro-inflammatory cytokine signaling [ 36 ]. These varied mechanisms demonstrate that this scaffold might be used for multiple drug discovery approaches. Research on the structure–activity relationship has consistently demonstrated that minor modifications to compounds can significantly impact their biological activity. Lipophilicity, electronic distribution, and steric bulk are among the factors that have a big effect on how well a drug acts and how it passes through the physiological membranes [ 37 ]. Combining coumarins with other pharmacophores has been a very successful technique to improve activity and get past resistance, but it may also make the molecules more complicated and hazardous [ 38 ]. Although coumarin derivatives reported to have promising biological potential, some derivatives have been linked to hepatotoxicity, fast metabolism, or low solubility in water, which can make it hard to use them in therapeutic settings [ 39 ]. These concerns underscore the imperative of amalgamating mechanistic insights and structure-activity relationship knowledge with predictive computational methodologies. In this context, AI-driven methods have the potential to systematically analyze large coumarin datasets, find good chemical patterns, and help design safer and more effective derivatives, as long as their predictions are based on solid chemical and biological principles [ 40 ]. From a medicinal chemistry perspective, coumarins are frequently regarded as a privileged scaffold capable of interacting with multiple classes of biological targets. The benzopyrone core provides a planar aromatic system with a conjugated lactone moiety that supports π–π stacking interactions, hydrogen bonding, and hydrophobic contacts within enzyme active sites. Substitution patterns around the coumarin nucleus play a decisive role in modulating biological selectivity. For example, electron-donating substituents at positions such as C-7 or C-8 can alter electron density across the aromatic system and influence interactions with catalytic residues, whereas bulky substituents or fused heterocycles may enhance target selectivity by occupying adjacent hydrophobic pockets [ 41 ]. These electronic and steric characteristics are particularly relevant for AI-based modeling because molecular descriptors used in machine-learning frameworks often encode properties such as electron distribution, lipophilicity, polar surface area, and steric volume. In addition, several coumarin derivatives are known to undergo metabolic transformation through cytochrome P450–mediated pathways, which may generate hydroxylated metabolites or contribute to hepatotoxic liabilities in certain chemical contexts. Incorporating such metabolic considerations into AI-driven ADMET prediction models enables early identification of potentially unstable or toxic coumarin derivatives and supports the rational optimization of pharmacokinetic profiles. Consequently, integrating medicinal chemistry knowledge with computational modeling frameworks allows AI systems to operate within chemically meaningful boundaries, ensuring that predictive algorithms remain grounded in the physicochemical realities of coumarin pharmacology [ 42 ]. 4. AI in target identification: principles and paradigms In recent years, the integration of AI into coumarin research has introduced a transformative dimension to drug discovery. AI-driven platforms, coupled with data-centric methodologies, have been increasingly applied to elucidate structure–activity relationships, predict binding affinities, and optimize lead generation. By automating complex analytical processes and reducing human bias, these computational tools enhance precision, efficiency, and innovation in therapeutic design [ 43 ]. Early applications of AI in coumarin-based drug discovery have already demonstrated their predictive power, with virtual screening and machine-learning models successfully identifying potential biological targets and guiding synthetic prioritization. Importantly, many of these AI-generated predictions have been validated through experimental studies, confirming the reliability and translational potential of such approaches [ 44 ]. Collectively, this synergy between AI and coumarin chemistry marks a pivotal step toward realizing the full therapeutic promise of this compound class, accelerating the journey from molecular design to clinical insight [ 45 ]. In practice, these computational strategies are typically implemented as part of a multi-step discovery pipeline involving data collection, target prioritization, virtual screening, lead optimization, and experimental validation. Beyond single-target prediction, AI-based target discovery increasingly incorporates disease network mapping to identify functionally connected biological pathways associated with pathological states. By integrating protein–protein interaction networks, signaling pathway databases, and multiomics datasets, machine-learning models can reconstruct disease-associated molecular networks and highlight key regulatory nodes susceptible to pharmacological modulation. Within this framework, coumarin derivatives can be evaluated not only for their affinity toward individual proteins but also for their potential to influence broader signaling circuits involved in disease progression. Such network-oriented analyses provide a systems-level perspective that improves target prioritization and helps identify mechanistic links between coumarin scaffolds and complex disease phenotypes [ 46 ]. Fundamental principles of AI and its analytical frameworks are redefining how therapeutic targets for coumarin derivatives are discovered and prioritized, as shown in Fig. 2 . Predictive modeling enables the direct identification of novel biological targets by integrating multidimensional datasets, while knowledge graph–based AI systems establish associative links between coumarin scaffolds and previously characterized molecular targets. In parallel, generative AI models leverage target–activity relationships to design new coumarin derivatives that are optimized for specific biological interactions in a single computational step [ 47 ]. Fig. 2. Open in a new tab Integrative multi-methodological framework for therapeutic target identification. Schematic representation of the three primary strategies used to identify and validate novel drug targets: (1) Multiomics (blue), encompassing genomics, epigenomics, transcriptomics, proteomics, and metabolomics for holistic biological profiling; (2) Experimental (purple), including chemical-genetic approaches (RNA, CRISPR/Cas9), affinity-based methods (photoaffinity, LD/AGC chemistry), and comparative profiling (stable-isotope labeling); and (3) Computational (green), featuring structure-based screening (pharmacophores, reverse docking) and AI-driven deep learning models. The bidirectional arrows indicate the synergistic feedback between data-driven discovery and laboratory validation. (For interpretation of the references to colour in this figure legend, the reader is referred to the Web version of this article.) By emphasizing data-centric exploration, AI transforms experimental limitations into valuable insights, allowing researchers to uncover hidden structure–function patterns within complex datasets. The integration of explainable prediction and ranking algorithms further enhances confidence in the interpretability and reproducibility of computational outcomes [ 48 ]. Moreover, the emerging concept of continuous learning loops introduces an adaptive, iterative workflow in which each analytical layer—predictive, associative, and generative—feeds into the next. This closed-loop system ensures that insights gained from one phase refine and strengthen the others, creating a dynamic and evolving strategy for AI-guided target identification and validation in coumarin-based drug discovery [ 49 ]. 5. From data to discovery: AI methodologies for coumarin targets AI methods used in coumarin-based drug discovery encompass several distinct methodological classes rather than a single uniform technology. These approaches differ in their computational architecture, input data requirements, and functional role within the discovery pipeline. Broadly, AI strategies can be categorized into machine-learning models for activity prediction, deep-learning architectures for complex pattern recognition, generative algorithms for molecular design, structure-based modeling frameworks for binding evaluation, and network-oriented approaches that integrate multiomics datasets to prioritize disease-relevant targets. Distinguishing these methodological classes helps clarify how AI contributes to different stages of coumarin discovery, ranging from early hypothesis generation and scaffold optimization to target validation and translational prioritization [ 50 ]. Table 1 summarizes representative AI approaches and their functional roles within coumarin-focused therapeutic discovery workflows. Table 1. Classification of AI methodologies applied to coumarin-based drug discovery. AI Approach Methodological Category Functional Role in Coumarin Discovery Example Application Ref. QSAR-based Machine Learning (Random Forest, SVM, Gradient Boosting) Ligand-based machine learning Predict biological activity of coumarin derivatives using molecular descriptors Activity prediction and lead prioritization [ 51 ] Graph Neural Networks (GNNs) Deep learning/structure-aware modeling Learn structural representations of coumarin scaffolds to guide scaffold optimization Prediction of structure–activity relationships and substituent effects [ 52 ] Molecular Docking with AI Rescoring Structure-based modeling Evaluate coumarin binding affinity toward protein targets using docking combined with ML-based scoring functions Binding validation and virtual screening [ 53 ] Generative Deep Learning Models (VAE, GAN, Transformer architectures) Generative chemistry Design novel coumarin analogues with optimized physicochemical or pharmacological properties De novo coumarin analogue generation [ 54 ] Reinforcement Learning Models Generative molecular optimization Iteratively optimize coumarin derivatives toward predefined objectives such as potency or ADMET properties Scaffold modification and lead optimization [ 40 ] Network Pharmacology and Knowledge Graph AI Systems biology modeling Identify coumarin–target–disease relationships across complex biological networks Polypharmacology prediction [ 55 ] Multiomics-integrated Machine Learning Systems-level AI analytics Integrate transcriptomic, proteomic, and metabolomic datasets to prioritize disease-relevant targets Target prioritization and mechanism discovery [ 56 ] AI-assisted ADMET Prediction Models Predictive modeling Forecast pharmacokinetic properties, toxicity risks, and metabolic liabilities of coumarin derivatives Early safety and drug-likeness optimization [ 37 ] Open in a new tab The integration of AI into coumarin research follows a structured drug discovery pipeline, spanning from initial data curation to translational prioritization as detailed in Table 2 . AI-driven approaches—such as network pharmacology for target identification and generative chemistry for lead optimization—enable the systematic refinement of coumarin scaffolds to enhance their therapeutic potential and ADMET profiles [ 57 ]. Table 2. AI-assisted discovery pipeline for coumarin-based therapeutics. Drug Discovery Stage Key Objective AI Tools/Computational Approaches Relevance for Coumarin Research Data acquisition and curation Compile chemical and biological datasets Text mining, database integration, knowledge graphs Aggregates coumarin bioactivity, structural, and pharmacological data from literature and public databases Target identification Identify disease-relevant biological targets Network pharmacology, multiomics ML, graph-based algorithms Links coumarin scaffolds to potential therapeutic targets across disease pathways Virtual screening Evaluate coumarin derivatives against candidate targets Molecular docking, ML-based scoring functions Prioritizes coumarin compounds with favorable predicted binding affinity Lead optimization Improve potency, selectivity, and physicochemical properties QSAR models, deep learning, generative chemistry Suggests structural modifications to coumarin scaffold to enhance activity ADMET prediction Predict pharmacokinetics and toxicity risks AI-assisted ADMET modeling, predictive ML classifiers Identifies metabolic liabilities and improves drug-like properties of coumarins Experimental validation Confirm biological activity and mechanism AI-guided assay prioritization, phenotype modeling Guides laboratory testing and mechanistic validation Translational prioritization Select candidates for further development Multi-criteria decision models, AI risk–benefit assessment Supports selection of coumarin candidates with favorable therapeutic potential Open in a new tab Comprehensive analyses of coumarin derivatives indicate that elucidating their structure–activity relationships should begin with large-scale data mining and systematic exploration of all documented molecular interactions. Extensive datasets available in scientific literature and public repositories provide valuable information on coumarin–target associations, which can be leveraged to train advanced molecular docking and deep-learning models for activity prediction [ 58 ]. These computational insights enable the rational design of new derivatives, which can subsequently be synthesized and evaluated through standardized biochemical assays or tailored phenotypic screening methods [ 59 ]. A risk–benefit assessment framework that integrates multiomics data—encompassing proteomics, transcriptomics, and metabolomics—alongside synthetic feasibility considerations, can effectively prioritize the most promising therapeutic targets [ 60 ]. Factors such as resource availability, development cost, and experimental turnaround time further guide decision-making during compound selection and validation. The implementation of carefully optimized in silico workflows—balancing rigorous virtual screening with flexible preclinical modeling—enhances predictive accuracy while minimizing both time and financial investment [ 61 ]. From a data-driven perspective, the growing body of literature on coumarins supports the strategic integration of AI tools to accelerate discovery. The coumarin scaffold, recognized for its structural diversity and favorable drug-like properties, serves as an adaptable platform for AI-based prediction and optimization [ 62 ]. Employing a template-driven modeling strategy, AI systems can map existing coumarin–bioactive target pairs across various diseases and use machine learning, molecular docking, and multisource data fusion to predict new interactions and refine risk–benefit models [ 63 ]. Continuous refinement of these predictive models remains essential, as even subtle conformational or electronic variations within the coumarin nucleus can significantly influence biological response, necessitating ongoing model calibration for reliable discovery outcomes [ 42 ]. AI-driven strategies for advancing target discovery in coumarin-based therapeutics can be categorized into four complementary approaches. The first relies on data mining to collect, refine, and interpret information from diverse public repositories, enabling the generation of actionable hypotheses regarding potential molecular targets [ 64 ]. The second employs molecular docking and predictive-binding simulations, which evaluate coumarin derivatives against disease-relevant proteins to estimate their binding affinity, selectivity, and mechanistic plausibility [ 65 ]. The third involves deep-learning frameworks that integrate multidimensional datasets to conduct risk–benefit assessments and prioritize targets according to their translational feasibility and biological significance [ 66 ]. Finally, the fourth approach utilizes multiomics data integration, combining genomic, proteomic, and metabolomic layers to construct context-specific biological maps that reveal how coumarins interact with complex cellular systems [ 67 ]. Embedding AI into the discovery pipeline for coumarin derivatives promotes a truly data-centric and adaptive research paradigm. When combined with knowledge graphs and predictive modeling, AI can generate new hypotheses and clarify underlying mechanisms of action [ 68 ]. Moreover, supervised and generative learning models establish continuous feedback loops, refining predictions over time and aligning computational outcomes with experimental observations. Through this synergy, AI enhances decision-making, optimizes resource use, and accelerates the translation of coumarin-based candidates from molecular insight to therapeutic innovation [ 69 ]. Recent advances in generative chemistry further extend the capabilities of AI-assisted molecular optimization. Deep generative models—including variational autoencoders, generative adversarial networks, and transformer-based architectures—can design novel coumarin analogues by learning chemical patterns from existing compound libraries and proposing new structures that satisfy predefined pharmacological or physicochemical constraints. Complementing these approaches, explainable AI (XAI) techniques are increasingly employed to interpret model predictions and identify structural features that drive biological activity. Methods such as feature attribution analysis, attention mapping, and interpretable descriptor modeling help reveal how specific substituents or electronic properties of the coumarin scaffold contribute to predicted activity profiles. Integrating generative design with explainable modeling can therefore guide medicinal chemists toward rational scaffold modification while maintaining transparency in algorithm-driven decision-making [ 70 ]. 5.1. Data mining and literature-driven insight Data mining has emerged as a cornerstone of modern drug discovery, particularly in the era of big data and AI. By systematically extracting and analyzing information from diverse biomedical databases, scientific literature, and experimental repositories, as illustrated in Fig. 3 , researchers can uncover hidden patterns and generate new hypotheses regarding drug–target interactions. In the context of coumarin research, data mining allows scientists to consolidate the extensive yet fragmented knowledge surrounding coumarin derivatives—their structural modifications, bioactivities, and pharmacological profiles—into cohesive, actionable datasets [ 71 ]. This integration of information provides a foundation for predictive modeling that supports the prioritization of potential therapeutic targets and informs the rational design of novel coumarin-based agents, with the aim of improving efficacy- and safety-related profiles [ 72 ]. Fig. 3. Open in a new tab Systematic data preprocessing pipeline for machine learning applications. This schematic illustrates the linear progression from Data Collection to Data Validation, highlighting the rigorous steps required to ensure data "readiness." In the context of AI-based drug discovery, this pipeline is essential for converting noisy, raw experimental data into high-fidelity inputs. By systematically addressing Data Cleaning and Transformation, the workflow minimizes algorithmic bias and enhances the predictive accuracy of deep learning models used to identify novel therapeutic candidates. Literature-driven insights complement these computational approaches by incorporating the contextual understanding that emerges from decades of experimental work on coumarin chemistry and pharmacology [ 38 ]. Manual and automated text-mining techniques can identify recurring molecular mechanisms, disease associations, and pharmacodynamic patterns across published studies [ 73 ]. For example, mining data from cancer, cardiovascular, and metabolic research has revealed consistent involvement of coumarin derivatives in modulating oxidative stress, inflammatory pathways, and enzyme inhibition—processes that underpin their therapeutic versatility [ 36 ]. The synthesis of such knowledge not only refines target prioritization but also enhances mechanistic understanding, guiding researchers toward more precise therapeutic hypotheses [ 74 ]. When combined, data mining and literature-based inference transform coumarin research from a largely empirical endeavor into a data-driven discovery ecosystem. These approaches facilitate cross-validation between experimental findings and computational predictions, reducing redundancy and accelerating innovation [ 75 ]. In therapeutic applications, this synergy has enabled the discovery of new coumarin scaffolds with multitarget potential, such as kinase inhibitors for oncology [ 76 ], aldose reductase inhibitors for diabetes [ 77 ], and acetylcholinesterase modulators for neurodegenerative diseases [ 78 ]. Ultimately, literature-informed data mining empowers scientists to bridge chemical diversity with biological relevance, ensuring that the next generation of coumarin therapeutics is guided by both evidence and insight [ 53 ]. 5.2. Molecular docking and predictive binding Coumarin scaffolds continue to attract considerable interest across multiple therapeutic domains, yet many of their molecular targets remain insufficiently characterized [ 79 ]. Molecular docking, as illustrated in Fig. 4 , has become an indispensable tool for prioritizing coumarin derivatives according to predicted binding affinity and selectivity toward disease-relevant proteins. For example, when a representative panel of five coumarin derivatives was docked and subsequently evaluated against acetylcholinesterase and butyrylcholinesterase, four compounds demonstrated inhibitory activity, with three exhibiting stronger inhibition than the reference drug galantamine. Only 4-substituted coumarin derivative lacked detectable activity. These observations were further supported by correlations between inhibitory potency and molecular size, as well as thermodynamic considerations consistent with Le Chatelier's principle [ 80 ]. Collectively, these findings underscore the value of docking-driven binding predictions in guiding the rational selection of coumarin frameworks for a wide range of pathological targets. Fig. 4. Open in a new tab Algorithm of molecular docking in the realm of drug discovery. Coumarin derivatives have also been explored as thiosemicarbazone-based candidates with promising activity against Mycobacterium tuberculosis and Trypanosoma cruzi , reflecting their potential in both bacterial and parasitic infections. Beyond their classical evaluation in tuberculosis models, a series of twenty-six coumarin–thiosemicarbazone hybrids has shown notable anticancer effects across several human tumor cell lines, along with activity against multidrug-resistant bacteria. Expanding the analytical toolkit, activity–molecular size mapping has provided additional insights into structure–activity drivers, including the reversible inhibition of acetylcholinesterase and butyrylcholinesterase observed for several compounds. A combined strategy—assessing both anti-tuberculosis properties and cholinesterase inhibition, alongside principal component-based unsupervised pattern recognition—has enabled a more comprehensive understanding of the biological space occupied by coumarin–thiosemicarbazone hybrids [ 81 ]. This integrative approach highlights their potential to operate along multiple axes of therapeutic relevance [ 82 ]. 5.3. Deep learning for target prioritization Deep learning has emerged as a transformative tool for prioritizing biological targets associated with coumarin derivatives, offering an analytical depth that surpasses traditional computational or statistical screening methods [ 83 ]. At its core, deep learning, as displayed in Fig. 5 , draws on multilayer neural networks capable of extracting subtle, non-linear patterns from complex biochemical datasets. When applied to coumarin research, these models integrate structure–activity relationships, molecular descriptors, pharmacophoric features, and multiomics signals to forecast which biological targets are most likely to respond to specific coumarin scaffolds [ 84 ]. This predictive power is particularly important for coumarins, whose diverse functional groups and modular architecture allow them to interact with a broad spectrum of proteins, enzymes, and signaling pathways [ 85 ]. Fig. 5. Open in a new tab Illustration of a deep learning algorithm, with a deep neural network serving as an example. A major advantage of deep learning in target prioritization lies in its ability to synthesize heterogeneous data streams—ranging from transcriptomics and proteomics to docking scores, phenotypic assays, and chemogenomic profiles—into a cohesive ranking framework. Instead of relying solely on binding affinity or known pathways, deep learning models evaluate a wider biological context, such as target essentiality, network connectivity, tissue specificity, and perturbation signatures. As a result, these systems can flag hidden or underexplored targets that may not emerge in classical screening but possess strong therapeutic potential for coumarin-derived compounds [ 86 ]. Such integrative assessments provide a more realistic picture of how coumarins may behave in living systems, enabling researchers to make evidence-driven decisions earlier in the discovery process [ 87 ]. In addition, deep learning models support iterative optimization by continuously updating predictions as new experimental data accumulate. High-throughput assays, cell-based phenotyping, and molecular dynamics simulations can be fed back into the network, allowing the model to refine its hypotheses and improve its predictive accuracy over time [ 88 ]. This dynamic, feedback-driven learning process mirrors how human experts refine their intuition but operates at a scale and speed beyond manual capability. For drug discovery teams working with coumarin scaffolds, such adaptability helps narrow down the most tractable and mechanistically promising targets with unprecedented efficiency [ 89 ]. While molecular docking remains a widely used approach for predicting ligand–protein interactions, its methodology differs substantially from data-driven machine-learning models. Docking techniques primarily rely on structural information of the target protein to estimate binding orientation and interaction energy between a ligand and the receptor binding site. These methods provide valuable insights into potential binding modes and molecular interactions but are often limited by simplified scoring functions and approximations of protein flexibility. In contrast, machine-learning models analyze large datasets of chemical structures and biological activity data to identify statistical patterns associated with ligand binding or pharmacological activity. Such models can capture complex nonlinear relationships between molecular descriptors and biological outcomes, enabling rapid screening of large chemical libraries without requiring explicit structural information for every target. However, machine-learning predictions are strongly dependent on the quality and diversity of the training data. Consequently, docking and machine-learning approaches are increasingly used in complementary workflows, where docking provides structural insight into ligand–target interactions while machine-learning models enhance predictive efficiency and prioritization of promising coumarin derivatives [ 90 ]. Ultimately, deep learning enriches the strategic landscape of coumarin-based therapeutic development by illuminating target hierarchies that balance potential efficacy, molecular tractability, and translational feasibility. By integrating multi-dimensional biological information, these models not only accelerate the early stages of discovery but also reduce downstream failure rates, guiding researchers toward coumarin–target interactions that hold the greatest promise for meaningful clinical impact [ 91 ]. If needed, these methods can be integrated with molecular docking, generative chemistry, or ADMET prediction pipelines to build a complete AI-enabled discovery ecosystem for coumarin derivatives [ 92 ]. Despite their predictive power, deep learning models present several important limitations that must be considered when applied to drug discovery tasks. These models often function as complex “black-box” systems, making it difficult to directly interpret which molecular features drive predicted biological activity. In addition, deep learning algorithms are highly dependent on the quality and representativeness of training datasets. If available datasets are biased toward particular chemical scaffolds or biological targets, the resulting models may inadvertently reproduce these biases and limit the generalizability of predictions. Consequently, the application of deep learning in coumarin-based drug discovery should be accompanied by careful dataset curation, model interpretability analyses, and experimental validation to ensure robust and reliable predictions [ 93 ]. 5.4. Multiomics integration for contextual targeting The adoption of multiomics strategies provides a powerful lens for understanding how coumarin derivatives operate within complex biological systems. Coumarins often engage multiple molecular pathways simultaneously, and their pharmacological effects cannot be fully explained by single-layer analyses. Multiomics integration, as shown on the left side of Fig. 2 , allows researchers to map these layered interactions, revealing how cellular states, metabolic fluxes, and regulatory networks shape coumarin responsiveness [ 94 ]. By contextualizing target biology within a system-wide framework, this approach supports more informed decisions about which pathways, tissues, or disease states are most susceptible to coumarin-based modulation [ 95 ]. An additional advantage of AI-based analytical frameworks lies in their ability to capture the inherently polypharmacological behavior of coumarin derivatives. Unlike highly selective synthetic ligands, many coumarins exert therapeutic effects through coordinated modulation of multiple biological targets. Network pharmacology approaches supported by AI can model these complex target constellations by integrating chemical–protein interaction datasets with disease-associated signaling networks. Through graph-based algorithms and knowledge-graph analysis, AI systems can identify clusters of interconnected proteins that may collectively contribute to disease progression and therefore represent coordinated intervention points for coumarin derivatives [ 96 ]. Such multi-target prediction strategies are particularly relevant for multifactorial diseases such as cancer, neurodegenerative disorders, and inflammatory conditions, where simultaneous modulation of several pathways may produce more durable therapeutic responses than single-target inhibition. Furthermore, AI-driven network analysis can assist in predicting potential synergistic effects between coumarin derivatives and other pharmacological agents, thereby supporting rational combination therapy design. By expanding analysis beyond single-target interactions, AI-enabled network pharmacology provides a more realistic framework for understanding the systems-level pharmacology of coumarin-based therapeutics [ 97 ]. In practice, multiomics data fusion uncovers molecular signatures that distinguish responsive from non-responsive phenotypes, enabling the identification of context-specific targets. For example, transcriptomic profiling may reveal coumarin-induced shifts in inflammation- or oxidative stress-related gene networks, while metabolomics can highlight downstream changes in redox metabolites, lipid mediators, or xenobiotic transformation pathways [ 98 ]. When these layers are combined with proteomic and phosphoproteomic data, researchers can pinpoint regulatory nodes—such as kinase cascades, redox-sensitive transcription factors, or metabolic checkpoints—that mediate or amplify coumarin activity. The resulting network-level insights create a clearer picture of how coumarin derivatives achieve therapeutic effects under different physiological or pathological conditions [ 99 ]. Multiomics integration also enhances target prioritization by identifying vulnerabilities that are unique to certain diseases. In cancers, for instance, coumarins may preferentially act on metabolic bottlenecks or DNA damage–response pathways revealed only through cross-omics comparison [ 100 ]. In inflammatory [ 101 ] or infectious diseases [ 102 ], immune-cell–specific epigenomic and transcriptomic patterns [ 103 ] may highlight coumarin-sensitive pathways governing cytokine production, pathogen clearance, or host defense. Such contextual targeting ensures that coumarin derivatives are matched not only to the appropriate biomolecular target, but also to the environment in which they are most likely to yield clinically meaningful outcomes. Furthermore, multiomics approaches help align chemical modifications of coumarin scaffolds with biological needs. Structure–activity relationships derived from metabolomics and proteomics can reveal how substituents influence metabolic stability, intracellular accumulation, or off-target interactions. These data-driven insights support the rational engineering of derivatives with the potential for enhanced selectivity and efficacy, while informing safety-related considerations [ 104 ]. Ultimately, integrating multiomics information transforms coumarin research from a trial-and-error process into a more precise, predictive, and context-aware strategy—accelerating the development of coumarin-based therapeutics with real translational potential [ 46 ]. Despite the potential advantages of multiomics integration for AI-assisted target discovery, several methodological challenges must be carefully addressed to ensure reliable results. Multiomics datasets are often generated using different experimental platforms, analytical pipelines, and sample preparation protocols, which may introduce systematic variability or batch effects. Consequently, data preprocessing steps such as normalization, batch-effect correction, and cross-platform harmonization are essential to ensure comparability between datasets before computational modeling is performed. In addition, rigorous validation strategies—including the use of independent datasets, cross-validation procedures, and experimental verification—are required to confirm that identified targets or pathway associations reflect biologically meaningful relationships rather than artifacts of data heterogeneity. Addressing these methodological considerations is critical for translating multiomics-driven AI predictions into robust insights for coumarin-based therapeutic discovery [ 105 ]. 6. AI-driven target validation: moving from in silico to in vivo The integration of AI-driven analytics into early drug discovery enhances both the speed and reliability of in vitro assays, while simultaneously reducing the likelihood of bias or confirmatory errors. When applied after a phenotypic screening campaign, AI helps to refine target identification by analyzing large compound libraries tested for observable biological effects. These phenotypic responses—validated across biochemical, genetic, and whole-organism systems including zebrafish and mammalian models—enable computational models to infer potential mechanisms of action, predict therapeutic efficacy, and anticipate off-target toxicities [ 106 ]. Rather than relying solely on classical pattern-recognition approaches, modern AI frameworks function as conceptual engines, generating mechanistic hypotheses and proposing experimental directions based on simulated system behavior. By modeling how different patient or biological subpopulations may respond to a compound, AI supports more informed decision-making and prioritization [ 107 ]. In practical drug discovery workflows, computational predictions generated by AI models must ultimately be translated into experimental validation steps to confirm biological relevance. Typically, AI-assisted screening first prioritizes a subset of coumarin derivatives predicted to exhibit favorable target interactions or pharmacological activity. These prioritized compounds are subsequently evaluated using in vitro biochemical or cell-based assays to verify predicted target engagement and biological effects. Positive results from these initial assays may then be followed by more detailed mechanistic studies, including enzyme inhibition kinetics, pathway analysis, or target knockdown experiments. Compounds demonstrating consistent activity across these stages may proceed to in vivo pharmacological evaluation and pharmacokinetic characterization. Such stepwise integration of computational prediction and experimental validation provides a practical framework for refining AI-generated hypotheses and improving the reliability of coumarin-based drug discovery pipelines [ 108 ]. For example, coumarin derivatives predicted through AI-assisted virtual screening to interact with specific enzymatic targets can first be evaluated using in vitro enzyme inhibition assays to confirm binding activity. In subsequent stages, active compounds may be tested in cell-based models to determine their effects on relevant signaling pathways or disease-associated phenotypes. For instance, coumarin candidates predicted to modulate oncogenic signaling pathways may be assessed in cancer cell lines to evaluate their impact on cell proliferation, apoptosis, or gene expression patterns. Compounds demonstrating consistent biological activity may then proceed to in vivo pharmacological studies to investigate therapeutic efficacy, pharmacokinetics, and toxicity profiles. Such sequential validation strategies provide a critical bridge between computational prediction and experimental confirmation, enabling AI-generated hypotheses to be translated into experimentally supported therapeutic insights [ 109 ]. An increasingly important concept in AI-assisted drug discovery is the establishment of iterative experimental–AI feedback loops. In this framework, computational predictions guide experimental assays, and the resulting biological data are continuously reintegrated into model training pipelines to refine predictive accuracy. Such closed-loop systems enable progressive improvement of structure–activity models and enhance the reliability of target prioritization strategies. Emerging concepts such as digital twin pharmacology further extend this paradigm by constructing computational representations of biological systems that simulate how individual patients or cellular environments may respond to specific compounds. Although still in early stages of development, these approaches illustrate how AI could eventually support dynamic modeling of coumarin pharmacology across multiple biological scales, from molecular interactions to organism-level responses [ 110 ]. Introducing AI-enabled risk-reduction strategies at the earliest stages of preclinical development significantly accelerates the identification of safe and effective drug candidates. Predicted target engagement, selectivity, and related pharmacological parameters can be validated early, allowing only the most promising molecules to advance. This streamlines the research process, minimizes unnecessary experimentation, and enhances collaboration among laboratories [ 111 ]. Importantly, AI can uncover functional relationships even in the absence of high homology data or fully characterized disease pathways, expanding the range of viable therapeutic hypotheses [ 112 ]. The performance of AI-based predictive models in ligand–protein interaction studies is typically evaluated using quantitative statistical metrics that reflect model accuracy and predictive reliability. In classification-based approaches, parameters such as accuracy, precision, recall, F1-score, and the receiver operating characteristic–area under the curve (ROC–AUC) are frequently used to assess the ability of machine-learning algorithms to correctly distinguish active from inactive compounds. In regression-based models that estimate binding affinities or activity values, additional metrics such as the coefficient of determination (R 2 ), mean squared error (MSE), and root mean squared error (RMSE) are commonly applied. These evaluation metrics provide an objective framework for comparing predictive performance across different computational models and datasets. Incorporating such quantitative assessments is essential to ensure that AI-driven predictions of coumarin–protein interactions are robust, reproducible, and informative for downstream experimental validation [ 113 ]. 6.1. Virtual screening to experimental validation Computationally generated target predictions can be efficiently translated into experimental validation when supported by strategic planning and collaboration with laboratories equipped to perform rapid bioassays. These in silico insights also provide a foundation for more focused experimental efforts, particularly when extended through molecular dynamics simulations that refine virtual screening outcomes [ 114 ]. By training next-generation convolutional neural networks, as illustrated in Fig. 6 , on large, well-curated datasets, researchers can begin to chart safer and faster preclinical pathways. Such models enable the construction of target-sensitive, high-risk cohorts that guide early “go/no-go” decisions and prioritize the most promising therapeutic directions [ 115 ]. Fig. 6. Open in a new tab Architecture of a convolutional neural network for ai-driven target prediction. AI can now be embedded directly into phenotypic screening workflows. Visualizing gene ontology terms throughout the screening cascade helps identify biological processes or pathways that repeatedly reappear among active hits, thereby enriching the context for linking genotype to phenotype. Integrating genotypic and metabolomic signatures with observed cellular responses strengthens the formulation of mechanistic hypotheses for each hit [ 116 ]. This multi-layered approach also facilitates the establishment of a robust baseline for inhibitor-induced phenotypes, enabling more balanced comparisons and ultimately improving confidence in the biological relevance of the final outcomes [ 117 ]. 6.2. AI in phenotypic and genotypic readouts Establishing the therapeutic value of a proposed treatment requires confirming that observed phenotypic changes truly arise from underlying functional genetic alterations [ 118 ]. Demonstrating these correlations provides an indirect yet powerful means of validating target engagement, especially when multiple complementary approaches—high-throughput screening, in vivo assays, and in silico analyses—converge on consistent findings [ 119 ]. For example, platforms such as TargetTSC and A-HeritageGP, each integrating three layers of omics data, have recently uncovered immunity-related molecular targets as part of a broad-spectrum anti-SARS-CoV-2 discovery effort. Several coumarin-based candidates highlighted in the second phase of the A-HeritageGP analysis emerged as plausible therapeutic leads. Their predicted loss-of-function associations support the hypothesized biological activity of the coumarin-rich decoction while posing minimal biosecurity concerns [ 120 ]. Beyond predictive target identification, recent advances emphasize the importance of AI-assisted functional validation frameworks capable of establishing causal relationships between coumarin exposure and biological response. One emerging strategy involves the integration of AI-guided CRISPR screening platforms, where machine-learning algorithms analyze genome-wide perturbation datasets to identify genes whose disruption alters cellular sensitivity to specific coumarin derivatives [ 121 ]. Such approaches enable the systematic identification of genetic dependencies and signaling pathways that mediate the observed pharmacological effects. Complementary transcriptomic perturbation mapping methods further enhance this validation process by comparing gene-expression signatures induced by coumarin compounds with large reference datasets derived from genetic knockdown or pharmacological perturbation experiments [ 122 ]. Through similarity-based analysis, these signatures can reveal mechanistic relationships between coumarin scaffolds and specific biological pathways. In parallel, causal network modeling and systems biology frameworks allow AI algorithms to integrate multiomics datasets into regulatory networks that capture gene–protein–metabolite interactions across cellular systems [ 123 ]. These models help distinguish direct target engagement from downstream secondary responses, thereby strengthening confidence in proposed mechanisms of action. By combining computational prediction with functional genomics and systems-level modeling, AI-driven pipelines can move beyond simple target nomination toward more rigorous validation of coumarin–target relationships [ 124 ]. Although these associations do not in themselves prove that the compounds directly modulate the identified targets, they serve as an important intermediate step in linking treatment exposure to specific molecular perturbations. A more rigorous demonstration of target dependency can be achieved through synthetic-lethality studies, in which disruption of a candidate target abolishes the therapeutic effect. Such experiments would clarify whether the target is essential for restoring immune homeostasis through coumarin-mediated mechanisms [ 125 ]. Additional validation using CRISPR–Cas9, as illustrated in Fig. 7 , would strengthen confidence in these mechanistic insights. Ultimately, integrating compositional databases enriched for metabolic and immune-activation pathways could enable more refined multiplex targeting strategies for coumarin-derived therapeutics [ 126 ]. Fig. 7. Open in a new tab Illustration of CRISPR–Cas9 gene editing algorithm. 6.3. Safer and faster preclinical pathways Simulation-based approaches in comparison with traditional trials, as shown in Fig. 8 , are increasingly recognized as essential tools for reducing risk in early-stage therapeutic development. Because biological systems are inherently complex and often unpredictable, simulations provide a practical way to explore how a candidate compound might behave before committing resources to live experiments [ 127 ]. Virtual experiments are not only more economical and faster to execute, but they also allow researchers to probe system-level responses that might otherwise be difficult or ethically challenging to study [ 128 ]. Although computational predictions can never fully replace empirical testing, scientifically grounded simulations contribute substantial value by informing hypotheses, highlighting mechanistic uncertainties, and helping to prioritize the most promising experimental directions. Their incorporation into study planning can help prioritize experimental efforts and guide early investigations toward more informative and safety-relevant endpoints [ 129 ]. Fig. 8. Open in a new tab The comparison of traditional methods and simulation algorithms in the drug discovery process. Adaptive designs complement these efforts by enabling modifications to the study protocol in real time, guided by interim data analyses. Such flexibility permits the refinement of dosing regimens, sample sizes, or endpoint selection while studies are still underway, ultimately improving efficiency without compromising scientific rigor [ 130 ]. Organizations that aim to accelerate development timelines and minimize risk stand to benefit significantly from integrating adaptive principles into their research strategy. As preclinical workflows increasingly adopt high-resolution quantitative data capture and advanced analytical platforms, and as computational models continue to evolve toward more precise predictions of drug action and outcome, the field is poised to develop preclinical programs that are not only faster and more cost-effective but also demonstrably safer and scientifically stronger [ 131 ]. 7. Case studies: coumarin-based therapeutics shaped by AI Concrete examples demonstrate how AI-driven methodologies are reshaping target identification and mechanistic discovery for coumarin derivatives. In 2021, an integrated literature-mining workflow identified more than 120 potential targets spanning diverse biological pathways. Subsequent virtual screening and experimental assays enabled the rapid recognition of a first-in-class antagonist of the oncogenic Myc–Max transcriptional complex, showcasing how AI accelerates both hit discovery and early biological validation [ 132 ]. That same year, coumarin–quinone hybrid molecules were rationally designed as selective inducers of reactive oxygen species in cancer cells. Incorporating a multi-parametric risk–benefit profile into a deep-learning model significantly improved target-ranking precision, and the development of a novel Sirt6 inhibitor provided a practical benchmark for evaluating model performance [ 133 ]. More recent investigations have leveraged multiomics platforms to clarify coumarin-mediated biological actions. In sepsis-induced cardiac dysfunction, combined cardiac proteomics and serum metabolomics highlighted proteins involved in calcium regulation as plausible therapeutic targets underlying the compound's cardioprotective effects. Complementary transcriptomic analyses in a membrane-mimetic model system further substantiated these mechanistic insights [ 134 ]. Collectively, these advances illustrate how the convergence of AI, cheminformatics, and multiomics technologies is expanding our understanding of coumarin pharmacology. By integrating trustworthy and mechanistically grounded AI pipelines, researchers are uncovering new therapeutic opportunities for this versatile class of compounds [ 135 ]. The practical application of AI in coumarin-based drug discovery spans from oncology target identification to generative molecular design, as outlined in Table 3 . These methodologies offer significant benefits, such as the rapid expansion of the coumarin chemical space and a deeper understanding of polypharmacology. However, the table also highlights critical limitations, including the risk of false positives in docking studies and the ongoing necessity for experimental validation of synthetic feasibility. Table 3. Representative examples of AI-assisted coumarin discovery. Application Area AI Methodology Coumarin-Based Strategy Reported Benefit Potential Limitation Ref. Oncology target discovery Literature mining + ML prediction Identification of coumarin derivatives interacting with oncogenic transcriptional complexes Accelerated identification of potential anticancer leads Requires experimental validation of predicted targets [ 76 ] Enzyme inhibition studies Molecular docking + ML scoring Prediction of coumarin inhibitors of enzymes such as acetylcholinesterase or kinases Rapid screening of large coumarin libraries Docking predictions may generate false positives [ 136 ] Multiomics-driven discovery Integrated transcriptomics and proteomics analysis Identification of pathways influenced by coumarin derivatives in disease models Improved mechanistic understanding of coumarin pharmacology Data integration challenges and variability between datasets [ 137 ] Generative molecular design Deep generative models and reinforcement learning Design of novel coumarin analogues with optimized physicochemical properties Expansion of chemical space beyond known derivatives Synthetic feasibility and experimental validation required [ 138 ] Polypharmacology modeling Network pharmacology and knowledge graphs Prediction of coumarin interactions with multiple disease-related targets Identification of multitarget therapeutic strategies Risk of overinterpreting network correlations [ 139 ] Open in a new tab One factor contributing to false-positive predictions in docking-based screening of coumarin derivatives is the highly planar and aromatic nature of the coumarin scaffold. The extended π-electron system of coumarins enables favorable π–π stacking interactions with aromatic amino acid residues such as phenylalanine, tyrosine, and tryptophan within protein binding pockets. Because many docking scoring functions heavily weight hydrophobic contacts and aromatic stacking interactions, planar coumarin molecules may appear to form energetically favorable binding poses even when the interactions are not sufficiently stable under physiological conditions. In addition, the relatively rigid structure of the coumarin core can geometrically fit into multiple binding sites, further increasing the likelihood of artificially favorable docking scores. Limitations in docking algorithms—such as simplified treatment of protein flexibility, solvent effects, and entropic contributions—may therefore amplify these apparent interactions and lead to false-positive predictions. Consequently, docking results involving planar coumarin derivatives should ideally be complemented with additional computational analyses or experimental validation to confirm genuine target engagement [ 140 ]. 8. Critical limitations, failure modes, and reproducibility challenges in AI-driven coumarin drug discovery Computer-based algorithms are playing an increasingly important role in the search for drugs based on coumarin. However, there are some significant challenges with these methods. Issues like limited data, incomplete datasets, and a focus on biologically active chemicals can make models less reliable and lead to overly optimistic conclusions [ 141 ]. The frequent absence of negative or unclear data worsens the problem of class imbalance, which increases the risk that AI models will pick up on quirks specific to the dataset rather than meaningful chemical-biological relationships [ 142 ]. Issues like data leakage, overfitting, and poor validation practices make things even more complicated, especially for machine learning and deep learning models that rely on molecular descriptors [ 143 ]. Using unnecessary features, splitting training and testing data improperly, or building models that are too complex can all make internal performance metrics look better than they really are, while hiding poor real-world predictive ability [ 144 ]. Structure-based approaches like molecular docking also have their limits, due to oversimplified scoring functions, rigid assumptions about proteins, and a higher rate of false positives in reverse docking—particularly for planar coumarin structures that tend to bind nonspecifically [ 145 ]. Beyond technical limitations, several conceptual failure modes may arise when AI models are applied to medicinal chemistry datasets. One concern involves the potential misinterpretation of structure–activity relationships when machine-learning models rely heavily on statistical correlations rather than mechanistically meaningful chemical features. In such cases, models may incorrectly attribute biological activity to structural fragments that merely co-occur in the training dataset rather than representing true pharmacophoric determinants [ 146 ]. A related challenge arises from the amplification of noisy or heterogeneous experimental datasets. Biological activity data derived from different laboratories, assay formats, or experimental conditions may contain inconsistencies that propagate through AI training pipelines, leading to unstable or misleading predictions. Additionally, AI systems trained predominantly on historically studied chemical scaffolds may inadvertently reinforce existing medicinal chemistry biases, favoring well-characterized structural motifs while underexploring novel chemical space. This phenomenon may limit the discovery of unconventional coumarin derivatives that fall outside traditional medicinal chemistry design rules. Recognizing these potential pitfalls is essential to ensure that AI-driven discovery remains guided by chemical reasoning, rigorous data curation, and continuous experimental validation [ 147 ]. Deep learning methods bring their own concerns around transparency, reproducibility, and practical use. Because these models often act like “black boxes,” it's harder to understand how they make decisions or to identify the best candidates. Differences in how datasets are prepared, models are built, or hyperparameters are chosen can lead to varying results across studies. On top of that, the lack of experimental validation for AI predictions raises the risk of overinterpreting findings and overlooking potential toxicity or pharmacokinetic issues [ 148 ]. These challenges necessitate the combining AI techniques with chemistry-driven, experimentally validated discovery processes to reliably advance coumarin-based therapies [ 44 ]. To address these challenges, several methodological strategies have been proposed to improve the robustness of AI models used in drug discovery. Techniques such as balanced sampling, synthetic data generation, and data augmentation can help reduce the impact of dataset imbalance by ensuring that both active and inactive compounds are adequately represented during model training. In addition, cross-validation approaches and independent test datasets are commonly employed to evaluate model generalizability and reduce the risk of overfitting. Ensemble modeling strategies, in which predictions from multiple algorithms are combined, may further enhance predictive stability. Implementing these mitigation strategies can significantly improve the reliability of AI-assisted predictions in coumarin-based therapeutic discovery [ 149 ]. 9. Ethical, regulatory, and practical considerations Ensuring that coumarin-based compounds progress smoothly toward clinical application requires rigorous attention to data integrity, transparency, and risk management [ 150 ]. High-quality, representative datasets are essential, as biased or poorly curated data can propagate errors throughout the AI models used in target discovery, drug screening, or mechanistic prediction. Data-centric development practices—such as careful dataset design, bias-aware augmentation strategies, and robust validation—help mitigate these concerns by revealing artifacts, hidden biases, and spurious correlations that may mislead decision-making [ 151 ]. Regulatory agencies, including the FDA and EMA, increasingly emphasize the need for interpretability and explainability, particularly when AI-derived insights influence preclinical or clinical outcomes [ 152 ]. From a translational perspective, the increasing use of AI-assisted molecular design also raises important regulatory and methodological questions. Regulatory agencies are increasingly evaluating how AI-derived insights should be documented, validated, and integrated into drug development workflows. Particular attention has been directed toward model interpretability, as decision-making algorithms used to prioritize compounds or targets must remain transparent enough to support regulatory review and scientific reproducibility. In parallel, dataset bias represents a persistent challenge in machine-learning-driven pharmacology. Many publicly available chemical and biological datasets are enriched for previously studied scaffolds or positive activity results, which can unintentionally bias predictive models toward historically favored chemical space while overlooking novel coumarin derivatives [ 70 ]. The broader reproducibility crisis observed in certain areas of computational drug discovery further highlights the need for standardized data curation practices, external validation datasets, and open reporting of model architectures and training procedures. Addressing these translational challenges is essential to ensure that AI-guided coumarin discovery remains scientifically robust and capable of supporting future clinical development efforts [ 153 ]. Understanding whether a model is being used for prediction, simulation, forecasting, or generative design is critical, as the proximity of the model to real-world decision points determines the level of interpretability required. In many cases, continuous monitoring of a model's real-world performance and proactive risk mitigation can be more impactful than overly exhaustive interrogation of in-sample behavior [ 154 ]. A responsible and transparent AI strategy can help address inherent uncertainties—through careful model selection, benchmarking, and stakeholder engagement—thereby supporting safety and reliability considerations. Given the emerging nature of AI-assisted coumarin research, maintaining caution during experimental validation remains essential [ 155 ]. Rapid in vitro assays may provide early indications of activity but often lack the depth required for confident translation without orthogonal confirmation. Phenotypic screening can efficiently identify promising candidates, yet it remains vulnerable to biological noise and false positives [ 156 ]. Novel or unexpected activities on untested targets highlight the need for continuous, real-time oversight of AI models to prevent over-interpretation. Furthermore, incorporating model-driven decisions back into training datasets must be performed judiciously to avoid reinforcing early-stage biases and compromising exploratory innovation [ 157 ]. Establishing a dynamic feedback loop—where predictions guide experiments, experimental results inform model refinement, and mechanistic insights are iteratively evaluated—can create a flexible system that aims to balance accelerated discovery with patient- and product-related safety considerations [ 158 ]. 10. Strategic roadmap for stakeholders Harnessing the full potential of AI in coumarin-based therapeutics requires coordinated, forward-thinking action across the entire research and development ecosystem. Academic scientists can lead this shift by integrating AI frameworks into early discovery and by articulating well-defined scientific and translational rationales for proposed targets. When presented with robust, de-risked biological hypotheses—especially in diseases with unmet medical need—industry partners are more likely to commit resources, accelerate development timelines, and pursue novel coumarin derivatives with confidence. Funding agencies should also expand the scope of AI-driven discovery within their grant portfolios, recognizing that computational target identification and validation are as essential to modern therapeutics as traditional high-throughput experimental screening. At the same time, clear governance standards addressing data quality, algorithmic fairness, and external validation will be crucial to maintaining scientific integrity and enabling responsible innovation. The AI-enabled discovery landscape is evolving rapidly, and coumarin-based agents appear particularly well suited to benefit from this progress. High-prevalence disease areas now offer a fertile environment in which AI tools can uncover actionable mechanisms, predict new target–compound interactions, and refine structure–activity understanding. With an expanding catalogue of mechanistic hypotheses, the emergence of powerful machine-learning approaches—such as gradient-boosted decision trees that enhance predictive accuracy—and increasingly comprehensive multiomics datasets, the field stands at a pivotal moment. Academic researchers, industry scientists, funding bodies, and policymakers all have an opportunity to capitalize on these favorable conditions and collectively drive the next generation of coumarin therapeutics. 11. Conclusion Target identification and validation continue to represent major rate-limiting steps in the development of coumarin-based therapeutics. Yet these obstacles can become strategic advantages when researchers and industry partners adopt AI to modernize and accelerate discovery workflows. Real-world case studies illustrate how AI is reshaping discovery: shifting efforts from purely algorithmic innovation toward empowering domain scientists; transitioning from one-off predictions to continuous, scalable learning; and training AI systems in the full therapeutic context rather than relying solely on abstract data patterns. Beyond its role as a computational tool, AI is increasingly reshaping the epistemological framework of natural product drug discovery by transforming how hypotheses are generated, tested, and refined across chemical and biological domains. Future efforts must focus on building such a strategy by developing robust data ecosystems—well-curated repositories, interoperable databases, and standardized ontologies—and integrating them with AI-driven processes that combine data-centric prediction models, genome-scale knowledge graphs, deep-learning architectures, and multi-omics fusion. Together, these tools enable the systematic generation, prioritization, and refinement of coumarin–target hypotheses that can be rapidly validated using experimental models. These emerging approaches mark a significant step in de-risking coumarin-focused discovery programs, facilitating faster hit identification, and improving the likelihood that early findings translate to human biology. Collectively, AI-enabled pipelines promise safer, more efficient, and substantially accelerated pathways for advancing coumarin-based drug candidates and shortening preclinical development timelines. Ethics approval Not applicable. Declaration of generative AI in scientific writing During the preparation of this work the author used Grammarly in order to assist with manuscript formatting and grammar checking. After using this service, the author reviewed and edited the content as needed and takes full responsibility for the content of the publication. Funding information This research received no specific grant from any funding agency in the public, commercial, or not-for-profit sectors. Declaration of competing interest The author declares that he has no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper. Acknowledgments The author would like to thank the University of Mosul, College of Pharmacy, for its support. BioRender.com was used to create the scientific figures included in this manuscript. Data availability Data sharing is not applicable to this article as no new datasets were generated or analyzed during the current study. References 1. Jia Z.-C., Yang X., Wu Y.-K., Li M., Das D., Chen M.-X., Wu J. The art of finding the right drug target: emerging methods and strategies. Pharmacol. Rev. 2024;76:896–914. doi: 10.1124/pharmrev.123.001028. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 2. Fahim Y.A., Hasani I.W., Kabba S., Ragab W.M. Artificial intelligence in healthcare and medicine: clinical applications, therapeutic advances, and future perspectives. Eur. J. Med. Res. 2025;30:848. doi: 10.1186/s40001-025-03196-w. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 3. Ratre P., Kulkarni S., Das S., Liang C., Mishra P.K., Thareja S. Medicinal chemistry aspects and synthetic strategies of coumarin as aromatase inhibitors: an overview. Med. Oncol. 2022;40:41. doi: 10.1007/s12032-022-01916-4. [ DOI ] [ PubMed ] [ Google Scholar ] 4. Lu Y., Huang W., Li Y., Xu Y., Wei Q., Sha C., Guo P. Leveraging artificial intelligence in antibody-drug conjugate development: from target identification to clinical translation in oncology. npj Precis. Oncol. 2025;9:374. doi: 10.1038/s41698-025-01159-2. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 5. Stefanachi A., Leonetti F., Pisani L., Catto M., Carotti A. Coumarin: a natural, privileged and versatile scaffold for bioactive compounds. Molecules. 2018;23:250. doi: 10.3390/molecules23020250. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 6. Zeki N.M., Mustafa Y.F. Harnessing artificial intelligence to discover the therapeutic potential of natural coumarins: a review study. Russ. J. Bioorg. Chem. 2025;51:1432–1452. doi: 10.1134/S1068162024607225. [ DOI ] [ Google Scholar ] 7. Paul D., Sanap G., Shenoy S., Kalyane D., Kalia K., Tekade R.K. Artificial intelligence in drug discovery and development. Drug Discov. Today. 2021;26:80–93. doi: 10.1016/j.drudis.2020.10.010. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 8. Gogoi B., Saikia S.P. Virtual screening and network pharmacology-based study to explore the pharmacological mechanism of clerodendrum species for anticancer treatment. Evid. Based Complement. Alternat. Med. 2022;2022:1–17. doi: 10.1155/2022/3106363. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 9. Li Y., Gu T., Yang C., Li M., Wang C., Yao L., Gu W., Sun D. AI-Assisted hypothesis generation to address challenges in cardiotoxicity research: simulation study using ChatGPT with GPT-4o. J. Med. Internet Res. 2025;27 doi: 10.2196/66161. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 10. Niazi S.K. Artificial intelligence in small-molecule drug discovery: a critical review of methods, applications, and real-world outcomes. Pharmaceuticals. 2025;18 doi: 10.3390/ph18091271. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 11. Citarella A., Vittorio S., Dank C., Ielo L. Syntheses, reactivity, and biological applications of coumarins. Front. Chem. 2024;12 doi: 10.3389/fchem.2024.1362992. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 12. Olatunde A., Tijjani H., Aremu A.O., Rauf A., Suleria H.A.R., Mubarak M.S. Editorial: chemistry, toxicity, synthesis, biological and pharmacological activities of coumarins and their derivatives: recent advances and future perspectives. Front. Pharmacol. 2024;15 doi: 10.3389/fphar.2024.1430985. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 13. Sinha S., Singh K., Ved A., Hasan S.M., Mujeeb S. Therapeutic journey and recent advances in the synthesis of coumarin derivatives. Mini Rev. Med. Chem. 2022;22:1314–1330. doi: 10.2174/1389557521666211116120823. [ DOI ] [ PubMed ] [ Google Scholar ] 14. Ramsis T.M., Ebrahim M.A., Fayed E.A. Synthetic coumarin derivatives with anticoagulation and antiplatelet aggregation inhibitory effects. Med. Chem. Res. 2023;32 doi: 10.1007/s00044-023-03148-1. [ DOI ] [ Google Scholar ] 15. Sayed M.T., Elsharabasy S.A., Abdel-Aziem A. Synthesis and antimicrobial activity of new series of thiazoles, pyridines and pyrazoles based on coumarin moiety. Sci. Rep. 2023;13:9912. doi: 10.1038/s41598-023-36705-0. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 16. Rawat A., Vijaya Bhaskar Reddy A. Recent advances on anticancer activity of coumarin derivatives. Eur. J. Med. Chem. Reports. 2022;5 doi: 10.1016/j.ejmcr.2022.100038. [ DOI ] [ Google Scholar ] 17. Jin A., Wang Y., Tong L., Liu G., Feng J., Li Y., Shen C., Wu W. Coumarins and flavones from Ficus erecta and their anti-inflammatory activity. J. Ethnopharmacol. 2024;333 doi: 10.1016/j.jep.2024.118472. [ DOI ] [ PubMed ] [ Google Scholar ] 18. Mishra P.S., Kumar A., Kaur K., Jaitak V. Recent developments in coumarin derivatives as neuroprotective agents. Curr. Med. Chem. 2024;31:5702–5738. doi: 10.2174/0929867331666230714160047. [ DOI ] [ PubMed ] [ Google Scholar ] 19. Sharapov A.D., Fatykhov R.F., Khalymbadzha I.A., Zyryanov G.V., Chupakhin O.N., Tsurkan M.V. Plant coumarins with Anti-HIV activity: isolation and mechanisms of action. Int. J. Mol. Sci. 2023;24 doi: 10.3390/ijms24032839. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 20. Giovannuzzi S., Hewitt C.S., Nocentini A., Capasso C., Flaherty D.P., Supuran C.T. Coumarins effectively inhibit bacterial Α-Carbonic anhydrases. J. Enzym. Inhib. Med. Chem. 2022;37:333–338. doi: 10.1080/14756366.2021.2012174. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 21. Zeki N.M., Mustafa Y.F. Annulated Heterocyclic[g]coumarin composites: synthetic approaches and bioactive profiling. Chem. Biodivers. 2024;21 doi: 10.1002/cbdv.202301855. [ DOI ] [ PubMed ] [ Google Scholar ] 22. Yildirim M., Poyraz S., Ersatir M. Recent advances on biologically active coumarin-based hybrid compounds. Med. Chem. Res. 2023;32:617–642. doi: 10.1007/s00044-023-03025-x. [ DOI ] [ Google Scholar ] 23. Anywar G., Muhumuza E. Bioactivity and toxicity of coumarins from African medicinal plants. Front. Pharmacol. 2024;14 doi: 10.3389/fphar.2023.1231006. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 24. Rastija V., Vrandečić K., Ćosić J., Kanižai Šarić G., Majić I., Karnaš M. Prospects of computer-aided molecular design of coumarins as ecotoxicologically safe plant protection agents. Appl. Sci. 2023;13:6535. doi: 10.3390/app13116535. [ DOI ] [ Google Scholar ] 25. Sadybekov A.V., Katritch V. Computational approaches streamlining drug discovery. Nature. 2023;616:673–685. doi: 10.1038/s41586-023-05905-z. [ DOI ] [ PubMed ] [ Google Scholar ] 26. Singh S., Kaushik N., Paliwal A., Sengar M.S., Paliwal D. Biological activity and therapeutic potential of coumarin derivatives: a comprehensive review. Curr. Drug Discov. Technol. 2025 doi: 10.2174/0115701638374546250323152450. [ DOI ] [ PubMed ] [ Google Scholar ] 27. Zengin Kurt B., Celebi G., Ozturk Civelek D., Angeli A., Akdemir A., Sonmez F., Supuran C.T. Tail-approach-based design and synthesis of coumarin-monoterpenes as carbonic anhydrase inhibitors and anticancer agents. ACS Omega. 2023;8:5787–5807. doi: 10.1021/acsomega.2c07459. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 28. Bester E., Petzer A., Petzer J.P. Coumarin derivatives as inhibitors of d-amino acid oxidase and monoamine oxidase. Bioorg. Chem. 2022;123 doi: 10.1016/j.bioorg.2022.105791. [ DOI ] [ PubMed ] [ Google Scholar ] 29. Gomaa M.S., Ali I.A.I., El Enany G., El Ashry E.S.H., El Rayes S.M., Fathalla W., Ahmed A.H.A., Abubshait S.A., Abubshait H.A., Nafie M.S. Facile synthesis of some coumarin derivatives and their cytotoxicity through VEGFR2 and topoisomerase II inhibition. Molecules. 2022;27:8279. doi: 10.3390/molecules27238279. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 30. Elkotamy M.S., Abdelrahman M.A., Eldehna W.M., Abdel-Aziz H.A., Abou-Seri S.M. Synthesis and biological evaluation of coumarin derivatives with promising anti-cancer activity: computational kinase profiling and molecular modeling analysis. Egypt. J. Chem. 2024 doi: 10.21608/ejchem.2024.343991.10974. 0–0. [ DOI ] [ Google Scholar ] 31. Kamel N.N., Aly H.F., Fouad G.I., Abd El-Karim S.S., Anwar M.M., Syam Y.M., Elseginy S.A., Ahmed K.A., Booles H.F., Shalaby M.B., Khalil W.K.B., Sandhir R., Deshwal S., Rizk M.Z. Anti-Alzheimer activity of new coumarin-based derivatives targeting acetylcholinesterase inhibition. RSC Adv. 2023;13:18496–18510. doi: 10.1039/D3RA02344C. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 32. Ibrar A., Shehzadi S.A., Saeed F., Khan I. Developing hybrid molecule therapeutics for diverse enzyme inhibitory action: active role of coumarin-based structural leads in drug discovery. Bioorg. Med. Chem. 2018;26:3731–3762. doi: 10.1016/j.bmc.2018.05.042. [ DOI ] [ PubMed ] [ Google Scholar ] 33. Kubrak T.P., Makuch-Kocka A., Aebisher D. Coumarins in anticancer therapy: mechanisms of action, potential applications and research perspectives. Pharmaceutics. 2025;17:595. doi: 10.3390/pharmaceutics17050595. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 34. Gadamsetti Srinandini, Kamala Govindarao, Prakash Degala Ravi, Prasanna V., Yaswini K., Srilaya A. A review of synthesis and therapeutic applications of coumarin derivatives. J. Pharma Insights Res. 2024;2:171–183. doi: 10.69613/23spvk43. [ DOI ] [ Google Scholar ] 35. Peter S., Sibali L.L. Recent developments on coumarin hybrids as antimicrobial agents. Antibiotics. 2025;14:1226. doi: 10.3390/antibiotics14121226. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 36. Saadati F., Modarresi Chahardehi A., Jamshidi N., Jamshidi N., Ghasemi D. Coumarin: a natural solution for alleviating inflammatory disorders. Curr. Res. Pharmacol. Drug Discov. 2024;7 doi: 10.1016/j.crphar.2024.100202. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 37. Zeki N.M., Mustafa Y.F. Digital alchemy: exploring the pharmacokinetic and toxicity profiles of selected coumarin-heterocycle hybrids. Results Chem. 2024;10 doi: 10.1016/j.rechem.2024.101754. [ DOI ] [ Google Scholar ] 38. Singh H., Singh J.V., Bhagat K., Gulati H.K., Sanduja M., Kumar N., Kinarivala N., Sharma S. Rational approaches, design strategies, structure activity relationship and mechanistic insights for therapeutic coumarin hybrids. Bioorg. Med. Chem. 2019;27:3477–3510. doi: 10.1016/j.bmc.2019.06.033. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 39. Pitaro M., Croce N., Gallo V., Arienzo A., Salvatore G., Antonini G. Coumarin-Induced hepatotoxicity: a narrative review. Molecules. 2022;27:9063. doi: 10.3390/molecules27249063. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 40. Li J., Lu Z., Wang L., Shi H., Chu B., Qu Y., Ye Z., Qu D. Novel coumarins derivatives for A. baumannii lung infection developed by high-throughput screening and reinforcement learning. J. Neuroimmune Pharmacol. 2024;19:32. doi: 10.1007/s11481-024-10134-w. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 41. Correa de Moraes M., Frassini R., Roesch-Ely M., Reisdorfer de Paula F., Barcellos T. Novel coumarin–nucleobase hybrids with potential anticancer activity: synthesis, in vitro cell-based evaluation, and molecular docking. Pharmaceuticals. 2024;17:956. doi: 10.3390/ph17070956. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 42. Hayat S., Wazzan S. A computational approach to predictive modeling using connection-based topological descriptors: applications in coumarin anti-cancer drug properties. Int. J. Mol. Sci. 2025;26:1827. doi: 10.3390/ijms26051827. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 43. Visan A.I., Negut I. Integrating artificial intelligence for drug discovery in the context of revolutionizing drug delivery. Life. 2024;14:233. doi: 10.3390/life14020233. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 44. Han R., Yoon H., Kim G., Lee H., Lee Y. Revolutionizing medicinal chemistry: the application of Artificial Intelligence (AI) in early drug discovery. Pharmaceuticals. 2023;16:1259. doi: 10.3390/ph16091259. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 45. Abbas M.K.G., Rassam A., Karamshahi F., Abunora R., Abouseada M. The role of AI in drug discovery. Chembiochem. 2024;25 doi: 10.1002/cbic.202300816. [ DOI ] [ PubMed ] [ Google Scholar ] 46. Yang L., Wang H., Zhu Z., Yang Y., Xiong Y., Cui X., Liu Y. Network pharmacology-driven sustainability: AI and multi-omics synergy for drug discovery in traditional Chinese medicine. Pharmaceuticals. 2025;18:1074. doi: 10.3390/ph18071074. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 47. Sarkar C., Das B., Rawat V.S., Wahlang J.B., Nongpiur A., Tiewsoh I., Lyngdoh N.M., Das D., Bidarolli M., Sony H.T. Artificial intelligence and machine learning technology driven modern drug discovery and development. Int. J. Mol. Sci. 2023;24:2026. doi: 10.3390/ijms24032026. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 48. Arsalan M.H., Mubin O., Al Mahmud A., Khan I.A., Hassan A.J. Mapping data-driven research impact science: the role of machine learning and artificial intelligence. Metrics. 2025;2:5. doi: 10.3390/metrics2020005. [ DOI ] [ Google Scholar ] 49. Johnson L., Devis D., Bacholer C., Leonard S.N. Closing the loop by expanding the scope: using learning analytics within a pragmatic adaptive engagement with complex learning environments. Front. Educ. 2024;9 doi: 10.3389/feduc.2024.1379520. [ DOI ] [ Google Scholar ] 50. Jha V., Holmelin F.L., Eriksson L.A. Binding analysis and structure-based design of tricyclic coumarin-derived MTHFD2 inhibitors as anticancer agents: insights from computational modeling. ACS Omega. 2023;8:14440–14458. doi: 10.1021/acsomega.2c08025. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 51. Kasmi R., Hadaji E., Chedadi O., El Aissouq A., Bouachrine M., Ouammou A. 2D-QSAR and docking study of a series of coumarin derivatives as inhibitors of CDK (anticancer activity) with an application of the molecular docking method. Heliyon. 2020;6 doi: 10.1016/j.heliyon.2020.e04514. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 52. Kim S., Han M., Park J., Lee K., Park S. Machine learning prediction of optical properties of coumarin derivatives using gaussian-weighted graph convolution and subgraph modular input. J. Chem. Inf. Model. 2025;65:4889–4897. doi: 10.1021/acs.jcim.5c00619. [ DOI ] [ PubMed ] [ Google Scholar ] 53. Abazari A., Sarrafi Y., Taheri P., Tajbakhsh M. Synthesis, biological evaluation, and molecular docking of novel coumarin-based bisimine derivatives. Sci. Rep. 2025;15 doi: 10.1038/s41598-025-26301-9. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 54. Zhang J., Chen H. De Novo Molecule Design Using Molecular Generative Models Constrained by Ligand–Protein Interactions. J. Chem. Inf. Model. 2022;62:3291–3306. doi: 10.1021/acs.jcim.2c00177. [ DOI ] [ PubMed ] [ Google Scholar ] 55. Dureshahwar K., Taufique M.S. Investigation and mechanism of coumarin for potential anti-epileptic targets using in-vitro SH-SY5Y cell line, molecular docking, and network pharmacology-based analysis. CNS Neurol. Disord. - Drug Targets. 2026;25:153–172. doi: 10.2174/0118715273340950250315071352. [ DOI ] [ PubMed ] [ Google Scholar ] 56. Zhang W., Jin Z., Huang R., Huang W., Li L., He Y., Zhou J., Tian C., Xiao L., Li P., Quan M., Zhang D., Du Q. Multi-omics analysis reveals genetic architecture and local adaptation of coumarins metabolites in populus. BMC Plant Biol. 2024;24:1170. doi: 10.1186/s12870-024-05894-9. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 57. Hu M., Yan H., Li H., Feng Y., Sun W., Ren Y., Ma L., Zeng W., Huang F., Jiang Z., Dong H. Use of network pharmacology and molecular docking to explore the mechanism of action of curcuma in the treatment of osteosarcoma. Sci. Rep. 2023;13:9569. doi: 10.1038/s41598-023-36687-z. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 58. Matsuzaka Y., Uesawa Y. A deep learning-based quantitative structure–activity relationship system construct prediction model of agonist and antagonist with high performance. Int. J. Mol. Sci. 2022;23:2141. doi: 10.3390/ijms23042141. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 59. Al‐Karmalawy A.A., Yousef T.A., Ali A.K., Zeidan M.A., Sharaky M., Al Khatib A.O., Ferjani H., Abbass E.M. Design and synthesis of new multi‐target coumarin analogues acting as carbonic anhydrases IX/XII and topoisomerase II inhibitors with apoptotic potential. Arch. Pharm. (Weinh.) 2025:358. doi: 10.1002/ardp.70094. [ DOI ] [ PubMed ] [ Google Scholar ] 60. Mustafa Y.F. Coumarins at the crossroads of drug development: enzyme modulation for clinical advances. Indian J. Clin. Biochem. 2025 doi: 10.1007/s12291-025-01331-2. [ DOI ] [ Google Scholar ] 61. Ru X., Xu L., Han W., Zou Q. In silico methods for drug-target interaction prediction. Cell Rep. Methods. 2025;5 doi: 10.1016/j.crmeth.2025.101184. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 62. Annunziata F., Pinna C., Dallavalle S., Tamborini L., Pinto A. An overview of coumarin as a versatile and readily accessible scaffold with broad-ranging biological activities. Int. J. Mol. Sci. 2020;21:1–83. doi: 10.3390/ijms21134618. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 63. Jasim M.H.M., Mustafa Y.F. Synthesis of acetaminophen‐based Coumarins as selective COX‐2 inhibitors: an in vitro‐in silico study. Chem. Biodivers. 2024;21 doi: 10.1002/cbdv.202401309. [ DOI ] [ PubMed ] [ Google Scholar ] 64. Manan A., Baek E., Ilyas S., Lee D. Digital alchemy: the rise of machine and deep learning in small-molecule drug discovery. Int. J. Mol. Sci. 2025;26:6807. doi: 10.3390/ijms26146807. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 65. Ghosh P., Singh R., Chatterjee C., Kumar A., Singh S.K. Computational screening of coumarin derivatives as inhibitors of the NACHT domain of NLRP3 inflammasome for the treatment of Alzheimer's disease. J. Biomol. Struct. Dyn. 2025;43:2187–2203. doi: 10.1080/07391102.2023.2294173. [ DOI ] [ PubMed ] [ Google Scholar ] 66. Abualigah L., Alomari S.A., Almomani M.H., Zitar R.A., Saleem K., Migdady H., Snasel V., Smerat A., Ezugwu A.E. Artificial intelligence-driven translational medicine: a machine learning framework for predicting disease outcomes and optimizing patient-centric care. J. Transl. Med. 2025;23:302. doi: 10.1186/s12967-025-06308-6. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 67. Monterisi S., Zuluaga M.Y.A., Senizza B., Cardarelli M., Rouphael Y., Colla G., Lucini L., Cesco S., Pii Y. The integrated multi-omics analysis unravels distinct roles of Malvaceae-derived protein hydrolysate and its molecular fraction in modulating tomato resilience under limited nitrogen availability. Plant Stress. 2025;15 doi: 10.1016/j.stress.2025.100771. [ DOI ] [ Google Scholar ] 68. Seal S., Mahale M., García-Ortegón M., Joshi C.K., Hosseini-Gerami L., Beatson A., Greenig M., Shekhar M., Patra A., Weis C., Mehrjou A., Badré A., Paisley B., Lowe R., Singh S., Shah F., Johannesson B., Williams D., Rouquie D., Clevert D.-A., Schwab P., Richmond N., Nicolaou C.A., Gonzalez R.J., Naven R., Schramm C., Vidler L.R., Mansouri K., Walters W.P., Wilk D.D., Spjuth O., Carpenter A.E., Bender A. Machine learning for toxicity prediction using chemical structures: pillars for success in the real world. Chem. Res. Toxicol. 2025;38:759–807. doi: 10.1021/acs.chemrestox.5c00033. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 69. Wan Z., Sun X., Li Y., Chu T., Hao X., Cao Y., Zhang P. Applications of artificial intelligence in drug repurposing. Adv. Sci. 2025;12 doi: 10.1002/advs.202411325. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 70. Lavecchia A. Explainable artificial intelligence in drug discovery: bridging predictive power and mechanistic insight. WIREs Comput. Mol. Sci. 2025;15 doi: 10.1002/wcms.70049. [ DOI ] [ Google Scholar ] 71. Tutone M., Almerico A.M. Computational strategies reshaping modern drug discovery. Molecules. 2026;31:200. doi: 10.3390/molecules31020200. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 72. Mazin Zeki N., Othman K.M.Z., Fakri Mustafa Y. Computational chemistry: a game-changer in the drug discovery field. Appl. Chem. Eng. 2025;8 doi: 10.59429/ace.v8i1.5601. ACE-5601. [ DOI ] [ Google Scholar ] 73. Palmblad M., Vleugels R., Bergquist J. Text mining and computational chemistry reveal trends in applications and applicability of capillary electrophoresis. TrAC, Trends Anal. Chem. 2023;159 doi: 10.1016/j.trac.2023.116946. [ DOI ] [ Google Scholar ] 74. Hacholli V.B., S M.R., P B.H., L M., P S., Kumar A., Szeleszczuk Ł., Gackowski M. Coumarin derivatives as anticancer agents: mechanistic landscape with an emphasis on breast cancer. Molecules. 2025;30:4167. doi: 10.3390/molecules30214167. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 75. Antonijević M.R., Avdović E.H., Simijonović D., Milanović Ž., Klisurić O.R., Alberto M.E., Russo N., Vojinović R., Marković Z.S. Comprehensive experimental and computational analysis of the structural and HSA binding properties of newly synthesized coumarin-trimethoxybenzohydrazide derivative. Polycycl. Aromat. Compd. 2024;44:7006–7030. doi: 10.1080/10406638.2024.2339846. [ DOI ] [ Google Scholar ] 76. Al Hasan M.S., Emon Y., Alshahrani M.Y., Mizan M., Uddin M.B., Hasan A.M.W., Sayeed M.A., Mia E., Yana N.T., Hossan R., Islam M.T. Coumarin derivatives as anticancer agents targeting PI3K-AKT-mTOR pathway: a comprehensive literature review. Med. Oncol. 2025;42:301. doi: 10.1007/s12032-025-02844-9. [ DOI ] [ PubMed ] [ Google Scholar ] 77. Waheed S.A., Mustafa Y.F. The in vitro effects of new albocarbon-based coumarins on blood glucose-controlling enzymes. J. Med. Chem. Sci. 2022;5:954–967. doi: 10.26655/JMCHEMSCI.2022.6.9. [ DOI ] [ Google Scholar ] 78. Zamanian M.Y., Parra R.M.R., Soltani A., Kujawska M., Mustafa Y.F., Raheem G., Al-Awsi L., Lafta H.A., Taheri N., Heidari M., Golmohammadi M., Bazmandegan G. Targeting Nrf2 signaling pathway and oxidative stress by resveratrol for Parkinson's disease: an overview and update on new developments. Mol. Biol. Rep. 2023;50:5455–5464. doi: 10.1007/s11033-023-08409-1. [ DOI ] [ PubMed ] [ Google Scholar ] 79. Bavishi A., Vala H., Thakrar S., Swami S., Sarkar D., Shukla R., Kamdar J., Shah A. Coumarin hybrids: dual-target candidates for future antimicrobial and antitubercular therapies. Future Med. Chem. 2025;17:1131–1142. doi: 10.1080/17568919.2025.2504331. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 80. Avelar M., Delgado-Acosta P.F., Guzmán-Ávila R., Flores-Morales V. Coumarin compounds against Alzheimer's disease: an in-silico approach. Results Chem. 2025;16 doi: 10.1016/j.rechem.2025.102481. [ DOI ] [ Google Scholar ] 81. Nunes J.A., Santos-Júnior P.F. da S., Gomes M.C., Ferreira L.A.S., Padilha E.K.A., Teixeira T.R., Stanger E.J., Kaur Y., da Silva E.B., Costa C.A.C.B., de Freitas J.D., de Araújo-Júnior J.X., Mendonça-Junior F.J.B., Giardini M.A., Siqueira-Neto J.L., Caffrey C.R., Zhan P., Cardoso S.H., da Silva-Júnior E.F. Nanomolar activity of coumarin-3-thiosemicarbazones targeting Trypanosoma cruzi cruzain and the T. brucei cathepsin L-like protease. Eur. J. Med. Chem. 2025;283 doi: 10.1016/j.ejmech.2024.117109. [ DOI ] [ PubMed ] [ Google Scholar ] 82. Zeki N.M., Mustafa Y.F. Coumarin hybrids: a sighting of their roles in drug targeting. Chem. Pap. 2024;78:5753–5772. doi: 10.1007/s11696-024-03498-z. [ DOI ] [ Google Scholar ] 83. Mulani A.O., Deshmukh M., Jadhav V., Chaudhari K., Mathew A.A., Salunkhe S. Transforming drug therapy with deep learning: the future of personalized medicine. Drug Res. 2025;75:326–333. doi: 10.1055/a-2682-5167. [ DOI ] [ PubMed ] [ Google Scholar ] 84. Kumar N., Srivastava R. Deep learning in structural bioinformatics: current applications and future perspectives. Briefings Bioinf. 2024;25 doi: 10.1093/bib/bbae042. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 85. Liu H., Chen Y., Cui F., Liao Y., Wang X. Synthesis, docking studies, biological activity of carbon monoxide release molecules based on coumarin derivatives. Front. Chem. 2022;10 doi: 10.3389/fchem.2022.996079. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 86. Debnath K., Rana P., Ghosh P. A survey on deep learning for drug-target binding prediction: models, benchmarks, evaluation, and case studies. Briefings Bioinf. 2025;26 doi: 10.1093/bib/bbaf491. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 87. Zeki N.M., Mustafa Y.F. Natural linear coumarin-heterocyclic conjugates: a review of their roles in phytotherapy. Fitoterapia. 2024;175 doi: 10.1016/j.fitote.2024.105929. [ DOI ] [ PubMed ] [ Google Scholar ] 88. Cheng Y., Bi X., Xu Y., Liu Y., Li J., Du G., Lv X., Liu L. Machine learning for metabolic pathway optimization: a review. Comput. Struct. Biotechnol. J. 2023;21:2381–2393. doi: 10.1016/j.csbj.2023.03.045. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 89. Jibroo R.N., Mustafa Y.F., Al-Shakarchi W. Synthesis and evaluation of linearly fused thiadiazolocoumarins as prospects with broad-spectrum bioactivity. Results Chem. 2024;7 doi: 10.1016/j.rechem.2024.101494. [ DOI ] [ Google Scholar ] 90. Li Y., Lin H., Yang H., Yuan Y., Zou R., Zhou G., Zhang L., Zheng H. Synergistic application of docking and machine learning for improved binding pose. Natl. Sci. Open. 2024 doi: 10.1360/nso/20230058. [ DOI ] [ Google Scholar ] 91. Rullo M., La Spada G., Stefanachi A., Macchia E., Pisani L., Leonetti F. Playing around the coumarin core in the discovery of multimodal compounds directed at alzheimer’s-related targets: a recent literature overview. Molecules. 2025;30:891. doi: 10.3390/molecules30040891. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 92. Zeki N.M., Mustafa Y.F. Novel heterocyclic coumarin annulates: synthesis and figuring their roles in biomedicine, bench-to-bedside investigation. Chem. Pap. 2024;78:4935–4951. doi: 10.1007/s11696-024-03441-2. [ DOI ] [ Google Scholar ] 93. Qadri Y.A., Shaikh S., Ahmad K., Choi I., Kim S.W., Vasilakos A.V. Explainable artificial intelligence: a perspective on drug discovery. Pharmaceutics. 2025;17:1119. doi: 10.3390/pharmaceutics17091119. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 94. Wörheide M.A., Krumsiek J., Kastenmüller G., Arnold M. Multi-omics integration in biomedical research – a metabolomics-centric review. Anal. Chim. Acta. 2021;1141:144–162. doi: 10.1016/j.aca.2020.10.038. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 95. Jiang W., Ye W., Tan X., Bao Y.-J. Network-based multi-omics integrative analysis methods in drug discovery: a systematic review. BioData Min. 2025;18:27. doi: 10.1186/s13040-025-00442-z. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 96. Huang C.-C., Chang K.-H., Chiu Y.-J., Chen Y.-R., Lung T.-H., Hsieh-Li H.M., Su M.-T., Sun Y.-C., Chen C.-M., Lin W., Lee-Chen G.-J. Multi-target effects of novel synthetic coumarin derivatives protecting Aβ-GFP SH-SY5Y cells against Aβ toxicity. Cells. 2021;10:3095. doi: 10.3390/cells10113095. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 97. Abbas S.R., Seol H., Abbas Z., Lee S.W. Exploring the role of artificial intelligence in smart healthcare: a capability and function-oriented review. Healthc. (Basel, Switzerland) 2025;13 doi: 10.3390/healthcare13141642. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 98. Cheng H., Liang M., Gao Y., Zhao W., Guo W.-F. Multiomics with evolutionary computation to identify molecular and module biomarkers for early diagnosis and treatment of complex disease. Genes (Basel) 2025;16:244. doi: 10.3390/genes16030244. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 99. Paliwal A., Jain S., Kumar S., Wal P., Khandai M., Khandige P.S., Sadananda V., Anwer M.K., Gulati M., Behl T., Srivastava S. Predictive modelling in pharmacokinetics: from in-silico simulations to personalized medicine. Expert Opin. Drug Metab. Toxicol. 2024;20:181–195. doi: 10.1080/17425255.2024.2330666. [ DOI ] [ PubMed ] [ Google Scholar ] 100. Dong M., Wang L., Hu N., Rao Y., Wang Z., Zhang Y. Integration of multi-omics approaches in exploring intra-tumoral heterogeneity. Cancer Cell Int. 2025;25:317. doi: 10.1186/s12935-025-03944-2. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 101. Younes A.H., Mustafa Y.F. Unveiling the biomedical applications of novel coumarins isolated from Capsicum Annuum L. seeds by a multivariate extraction technique. Chem. Biodivers. 2024;21 doi: 10.1002/cbdv.202400581. [ DOI ] [ PubMed ] [ Google Scholar ] 102. Cheke R.S., Patel H.M., Patil V.M., Ansari I.A., Ambhore J.P., Shinde S.D., Kadri A., Snoussi M., Adnan M., Kharkar P.S., Pasupuleti V.R., Deshmukh P.K. Molecular insights into coumarin analogues as antimicrobial agents: recent developments in drug discovery. Antibiotics. 2022;11:566. doi: 10.3390/antibiotics11050566. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 103. Mustafa Y.F. Natural immune boosters: coumarins as emerging potentiators of immunity in drug discovery. Phytomed. Plus. 2025;5 doi: 10.1016/j.phyplu.2025.100880. [ DOI ] [ Google Scholar ] 104. Gharios R., Francis R.M., DeForest C.A. Chemical and biological engineering strategies to make and modify next-generation hydrogel biomaterials. Matter. 2023;6:4195–4244. doi: 10.1016/j.matt.2023.10.012. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 105. Ivanisevic T., Sewduth R.N. Multi-omics integration for the design of novel therapies and the identification of novel biomarkers. Proteomes. 2023;11 doi: 10.3390/proteomes11040034. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 106. Niazi S.K. The coming of age of AI/ML in drug discovery, development, clinical testing, and manufacturing: the FDA perspectives. Drug Des. Dev. Ther. 2023;17:2691–2725. doi: 10.2147/DDDT.S424991. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 107. Dharmasivam M., Kaya B., Akinware A., Azad M.G., Richardson D.R. Leading artificial intelligence–driven drug discovery platforms: 2025 landscape and global outlook. Pharmacol. Rev. 2026;78 doi: 10.1016/j.pharmr.2025.100102. [ DOI ] [ PubMed ] [ Google Scholar ] 108. Mohammed Alshaher M., Fakri Mustafa Y. From laboratory to computer models: enhancing coumarin discovery through interdisciplinary research. Appl. Chem. Eng. 2025;8:5613. doi: 10.59429/ace.v8i1.5613. [ DOI ] [ Google Scholar ] 109. Kishk S.M., Eltamany E.E., Nafie M.S., Khinkar R.M., Hareeri R.H., Elhady S.S., Yassen A.S.A. Design and synthesis of coumarin derivatives as cytotoxic agents through PI3K/AKT signaling pathway inhibition in HL60 and HepG2 cancer cells. Molecules. 2022;27:6709. doi: 10.3390/molecules27196709. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 110. Vt I., Asireddy S.R., Vallarapu N.K., Madhira J., T R. Digital twins in drug discovery: a paradigm shift shaping pharmaceutical innovation. Int. J. Pharm. Sci. Nanotechnology(IJPSN) 2024;17:7628–7637. doi: 10.37285/ijpsn.2024.17.5.9. [ DOI ] [ Google Scholar ] 111. Fu C., Chen Q. The future of pharmaceuticals: artificial intelligence in drug discovery and development. J. Pharm. Anal. 2025 doi: 10.1016/j.jpha.2025.101248. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 112. Mustafa Y.F. 3-mercaptocoumarins as potential bioactive candidates: from novel synthesis to comparative analysis. J. Mol. Struct. 2025;1320 doi: 10.1016/j.molstruc.2024.139657. [ DOI ] [ Google Scholar ] 113. Harren T., Gutermuth T., Grebner C., Hessler G., Rarey M. Modern machine‐learning for binding affinity estimation of protein–ligand complexes: progress, opportunities, and challenges. WIREs Comput. Mol. Sci. 2024;14 doi: 10.1002/wcms.1716. [ DOI ] [ Google Scholar ] 114. Wu Y., Chen Z., Chen X., Li C., Zengin G., Li M.-Y. Computational strategies for antimicrobial discovery: from machine learning to multiscale simulation. Curr. Mol. Pharmacol. 2025;18:63–82. doi: 10.1016/j.cmp.2025.10.003. [ DOI ] [ Google Scholar ] 115. Bhattacharya M., Lo Y.-H., Chatterjee S., Das A., Wen Z.-H., Chakraborty C. Deep learning in next-generation vaccine development for infectious diseases. Mol. Ther. Nucleic Acids. 2025;36 doi: 10.1016/j.omtn.2025.102586. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 116. Cheng C., Liu Y., Sun L., Fan J., Sun X., Zheng J.-S., Zheng L., Zhu Y., Zhou D. Integrative metabolomics and genomics reveal molecular signatures for type 2 diabetes and its cardiovascular complications. Cardiovasc. Diabetol. 2025;24:166. doi: 10.1186/s12933-025-02718-4. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 117. Mustafa Y.F. Synthesis of 7,8-dihydroxy-4-phenylbenzo[g]coumarins as potential multitarget anti-skin-aging candidates. J. Mol. Struct. 2025;1321 doi: 10.1016/j.molstruc.2024.139806. [ DOI ] [ Google Scholar ] 118. Ghasemi Darestani N., Gilmanova A.I., Al-Gazally M.E., Zekiy A.O., Ansari M.J., Zabibah R.S., Jawad M.A., Al-Shalah S.A.J., Rizaev J.A., Alnassar Y.S., Mohammed N.M., Mustafa Y.F., Darvishi M., Akhavan-Sigari R. Mesenchymal stem cell-released oncolytic virus: an innovative strategy for cancer treatment. Cell Commun. Signal. 2023;21:43. doi: 10.1186/s12964-022-01012-0. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 119. Kasim S.M., Abdulaziz N.T., Jasim M.H., Mustafa Y.F. Resveratrol in cancer chemotherapy: is it a preventer, protector, or fighter? Eurasian Chem. Commun. 2023;5:576–587. doi: 10.22034/ecc.2023.379480.1586. [ DOI ] [ Google Scholar ] 120. Rudiansyah M., Jasim S.A., Mohammad pour Z.G., Athar S.S., Jeda A.S., Doewes R.I., Jalil A.T., Bokov D.O., Mustafa Y.F., Noroozbeygi M., Karampoor S., Mirzaei R. Coronavirus disease 2019 (COVID‐19) update: from metabolic reprogramming to immunometabolism. J. Med. Virol. 2022;94:4611–4627. doi: 10.1002/jmv.27929. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 121. Mustafa Y.F. Applications of artificial intelligence in the synthesis, docking, and pharmacological profiling of coumarins. Appl. Chem. Eng. 2025;8:5678. doi: 10.59429/ace.v8i2.5678. [ DOI ] [ Google Scholar ] 122. Keskar V., Shenoy A., Desai S. AI-based prediction of drug-gene interactions modulating tight junction integrity: a deep learning framework highlighting multiple therapeutic targets. J. Oral Biol. Craniofac. Res. 2025;15:1786–1792. doi: 10.1016/j.jobcr.2025.10.014. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 123. Li X., Xing J., Zhang S., Zhou J. Editorial: advancing drug discovery with AI: drug-target interactions, mechanisms of action, and screening. Front. Pharmacol. 2025;16 doi: 10.3389/fphar.2025.1721323. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 124. Abdollahzadeh Jamalabadi M.Y. 2025. A Comprehensive Review on Deep Learning for Genomics and AI in Drug Discovery. [ DOI ] [ Google Scholar ] 125. Mustafa Y.F. Synthesis of novel 6-aminocoumarin derivatives as potential –biocompatible antimicrobial and anticancer agents. J. Mol. Struct. 2025;1320 doi: 10.1016/j.molstruc.2024.139658. [ DOI ] [ Google Scholar ] 126. Abdul-Hussin I.F., Alkhalidi M.H.O., Al-Musawi S., Alshalah L.A.M., Sheykhhasan M. CRISPR-Cas9 in functional genomics: implications for target validation in precision oncology. Trends Pharm. Biotechnol. 2025;3:36–48. doi: 10.57238/tpb.2025.153196.1026. [ DOI ] [ Google Scholar ] 127. Ismael R.N., Mustafa Y.F., Al-qazaz H.K. Citrullus lanatus, a potential source of medicinal products : a review. J. Med. Chem. Sci. 2022;5:607–618. doi: 10.26655/JMCHEMSCI.2022.4.16. [ DOI ] [ Google Scholar ] 128. Li J., Liang W. Effectiveness of virtual laboratory in engineering education: a meta-analysis. PLoS One. 2024;19 doi: 10.1371/journal.pone.0316269. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 129. Procopio A., Cesarelli G., Donisi L., Merola A., Amato F., Cosentino C. Combined mechanistic modeling and machine-learning approaches in systems biology – a systematic literature review. Comput. Methods Progr. Biomed. 2023;240 doi: 10.1016/j.cmpb.2023.107681. [ DOI ] [ PubMed ] [ Google Scholar ] 130. Pallmann P., Bedding A.W., Choodari-Oskooei B., Dimairo M., Flight L., Hampson L.V., Holmes J., Mander A.P., Odondi L., Sydes M.R., Villar S.S., Wason J.M.S., Weir C.J., Wheeler G.M., Yap C., Jaki T. Adaptive designs in clinical trials: why use them, and how to run and report them. BMC Med. 2018;16:29. doi: 10.1186/s12916-018-1017-7. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 131. Alasmari M.S., Albusaysi S., Elhefnawy M., Ali A.M., Altigani K., Almoslem M., Alharbi M., Alghamdi J., Alsultan A. Model-informed drug discovery and development approaches to inform clinical trial design and regulatory decisions: a primer for the MENA region. Saudi Pharm. J. 2024;32 doi: 10.1016/j.jsps.2024.102207. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 132. Singh A., Kumar A., Kumar P., Nayak N., Bhardwaj T., Giri R., Garg N. A novel inhibitor L755507 efficiently blocks c-Myc–MAX heterodimerization and induces apoptosis in cancer cells. J. Biol. Chem. 2021;297 doi: 10.1016/j.jbc.2021.100903. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 133. Pangal A., Ahmed K. Synthesis and biological evaluation of coumarin-quinone hybrids as multifunctional bioactive agents. ADMET DMPK. 2022;11:81–96. doi: 10.5599/admet.1468. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 134. Yu X., Gao J., Zhang C. Sepsis-induced cardiac dysfunction: Mitochondria and energy metabolism. Intensive Care Med. Exp. 2025;13:20. doi: 10.1186/s40635-025-00728-w. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 135. Keramida P., Syrigos N.K., Kouvela M., Poulakou G., Charpidou A., Fiste O. AΙ-Driven drug repurposing: applications and challenges. Medicines. 2025;12:28. doi: 10.3390/medicines12040028. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 136. Yusufzai S.K., Khan M.S., Sulaiman O., Osman H., Lamjin D.N. Molecular docking studies of coumarin hybrids as potential acetylcholinesterase, butyrylcholinesterase, monoamine oxidase A/B and β-amyloid inhibitors for Alzheimer's disease. Chem. Cent. J. 2018;12:128. doi: 10.1186/s13065-018-0497-z. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 137. Chen S., Shen L., Zeng Y., Feng S., Luo H., Wang G. Integrated metabolomic and transcriptomic analyses reveal the coumarin biosynthesis pathway and key regulatory genes in the pericarp of Zanthoxylum. Plants. 2026;15:769. doi: 10.3390/plants15050769. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 138. Naz N., Nazli Z.-H., Parveen S., Shafiq N. De Novo Drug Design, Synthesis, Biological Evaluation, and Structural Examination of Novel Coumarin-Based Pyrimidine Co-Drugs Accompanied by Molecular Docking and DFT Studies. ACS Omega. 2026;11:15233–15253. doi: 10.1021/acsomega.5c12395. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 139. Li Z., Li Q. Study on the anti-inflammatory mechanism of coumarins in Peucedanum decursivum based on spatial metabolomics combined with network pharmacology. Molecules. 2024;29:3346. doi: 10.3390/molecules29143346. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 140. Pinzi L., Rastelli G. Molecular docking: shifting paradigms in drug discovery. Int. J. Mol. Sci. 2019;20:4331. doi: 10.3390/ijms20184331. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 141. Mihai D.P., Nitulescu G.M. Computer-aided drug design and drug discovery. Pharmaceuticals. 2025;18:436. doi: 10.3390/ph18030436. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 142. Rodrigues T. The good, the bad, and the ugly in chemical and biological data for machine learning. Drug Discov. Today Technol. 2019;32–33:3–8. doi: 10.1016/j.ddtec.2020.07.001. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 143. Kapoor S., Narayanan A. Leakage and the reproducibility crisis in machine-learning-based science. Patterns (New York, N.Y.) 2023;4 doi: 10.1016/j.patter.2023.100804. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 144. Han H. Challenges of reproducible AI in biomedical data science. BMC Med. Genom. 2025;18:8. doi: 10.1186/s12920-024-02072-6. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 145. Grinter S.Z., Zou X. Challenges, applications, and recent advances of protein-ligand docking in structure-based drug design. Molecules. 2014;19:10150–10176. doi: 10.3390/molecules190710150. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 146. Gangwal A., Lavecchia A. Explainable AI methods for drug discovery: a survey of interpretability, metrics and mechanistic insight. Comput. Sci. Rev. 2026;61 doi: 10.1016/j.cosrev.2026.100943. [ DOI ] [ Google Scholar ] 147. Kant S., Deepika, Roy S. Artificial intelligence in drug discovery and development: transforming challenges into opportunities. Discov. Pharm. Sci. 2025;1:7. doi: 10.1007/s44395-025-00007-3. [ DOI ] [ Google Scholar ] 148. Serrano D.R., Luciano F.C., Anaya B.J., Ongoren B., Kara A., Molina G., Ramirez B.I., Sánchez-Guirales S.A., Simon J.A., Tomietto G., Rapti C., Ruiz H.K., Rawat S., Kumar D., Lalatsa A. Artificial Intelligence (AI) applications in drug discovery and drug delivery: revolutionizing personalized medicine. Pharmaceutics. 2024;16:1328. doi: 10.3390/pharmaceutics16101328. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 149. Kannan M., Umamaheswari D., Manimekala B., Mary I.P.S., Savitha P.M., Rozario J. An enhancement of machine learning model performance in disease prediction with synthetic data generation. Sci. Rep. 2025;15 doi: 10.1038/s41598-025-15019-3. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 150. Zeki M.N., Mustafa Y.F. Synthesis and evaluation of novel ring-conjugated coumarins as biosafe broad-spectrum antimicrobial candidates. J. Mol. Struct. 2024;1309 doi: 10.1016/j.molstruc.2024.138192. [ DOI ] [ Google Scholar ] 151. Wu Y., Wang N., Xiong P., Wang R., Deng J., Ouyang D. Artificial intelligence for drug delivery: yesterday, today and tomorrow. Acta Pharm. Sin. B. 2025 doi: 10.1016/j.apsb.2025.09.022. [ DOI ] [ Google Scholar ] 152. Ferreira F.J.N., Carneiro A.S. AI-Driven drug discovery: a comprehensive review. ACS Omega. 2025;10:23889–23903. doi: 10.1021/acsomega.5c00549. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 153. Sofi F.A. Coumarin as a privileged natural product scaffold towards the development of antibacterial agents targeting ESKAPE pathogens. Eur. J. Med. Chem. 2026;309 doi: 10.1016/j.ejmech.2026.118735. [ DOI ] [ PubMed ] [ Google Scholar ] 154. Jibroo R.N., Mustafa Y.F., Al-Shakarchi W. Heterocycles fused on a 6,7-coumarin framework: an in-depth review of their structural and pharmacological diversity. Chem. Pap. 2024;78:7239–7311. doi: 10.1007/s11696-024-03613-0. [ DOI ] [ Google Scholar ] 155. Andrikyan W., Sametinger S.M., Kosfeld F., Jung-Poppe L., Fromm M.F., Maas R., Nicolaus H.F. Artificial intelligence-powered chatbots in search engines: a cross-sectional study on the quality and risks of drug information for patients. BMJ Qual. Saf. 2025;34:100–109. doi: 10.1136/bmjqs-2024-017476. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 156. Alshaher M.M., Mustafa Y.F. Linear pyranocoumarins are potential dazzling dancers between nature, chemistry, and clinical application. Phytomed. Plus. 2025;5 doi: 10.1016/j.phyplu.2025.100785. [ DOI ] [ Google Scholar ] 157. Mittermaier M., Raza M.M., Kvedar J.C. Bias in AI-based models for medical applications: challenges and mitigation strategies. npj Digit. Med. 2023;6:113. doi: 10.1038/s41746-023-00858-z. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 158. Ijjeh Y., Alsarayreh N., Rifai A., Abdelnabi H., Al-Mahamid S., Alqudah D.A., Nsairat H., Lafi Z., Mousa M.A., Al-Hiari Y., Alshaer W. Quality by digital design for accelerated sustainable nanomedicine development. Eur. J. Pharmaceut. Sci. 2025;213 doi: 10.1016/j.ejps.2025.107239. [ DOI ] [ PubMed ] [ Google Scholar ] Associated Data This section collects any data citations, data availability statements, or supplementary materials included in this article. Data Availability Statement Data sharing is not applicable to this article as no new datasets were generated or analyzed during the current study. Articles from Pharmaceutical Science Advances are provided here courtesy of Elsevier ACTIONS View on publisher site PDF (4.5 MB) Cite Collections Permalink PERMALINK Copy RESOURCES Similar articles Cited by other articles Links to NCBI Databases Cite Copy Download .nbib .nbib Format: AMA APA MLA NLM Add to Collections Create a new collection Add to an existing collection Name your collection * Choose a collection Unable to load your collection due to an error Please try again Add Cancel Follow NCBI NCBI on X (formerly known as Twitter) NCBI on Facebook NCBI on LinkedIn NCBI on GitHub NCBI RSS feed Connect with NLM NLM on X (formerly known as Twitter) NLM on Facebook NLM on YouTube National Library of Medicine 8600 Rockville Pike Bethesda, MD 20894 Web Policies FOIA HHS Vulnerability Disclosure Help Accessibility Careers NLM NIH HHS USA.gov Back to Top