The Application of Artificial Intelligence in Public Health Surveillance in Portugal: An Exploratory Study of Expert Perspectives.

The Application of Artificial Intelligence in Public Health Surveillance in Portugal: An Exploratory Study of Expert Perspectives - PMC Skip to main content An official website of the United States government Here's how you know Here's how you know Search Log in Primary site navigation Search Log in Search PMC Full-Text Archive Search in PMC Journal List User Guide 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 Port J Public Health . 2026 Jan 29;44(2):120–139. doi: 10.1159/000550578 Show available content in en pt The Application of Artificial Intelligence in Public Health Surveillance in Portugal: An Exploratory Study of Expert Perspectives Liliana Certo a, b, ✉, ✉ , Maurício Alves c , Teresa Magalhães d Author information Article notes Copyright and License information a National School of Public Health, NOVA University Lisbon, Lisbon, Portugal b Tagus Estuary Local Health Unit, Vila Franca de Xira, Portugal c Alto Alentejo Local Health Unit, Lisbon, Portugal d NOVA National School of Public Health, Public Health Research Centre, Comprehensive Health Research Center, CHRC, REAL, CCAL, NOVA University Lisbon, Lisbon, Portugal ✉ Correspondence to: Liliana Certo, lrf.certo@ensp.unl.pt ✉ Corresponding author. Received 2025 Jul 27; Accepted 2026 Jan 5; Collection date 2026 Oct. © 2026 The Author(s). Published by S. Karger AG, Basel on behalf of NOVA National School of Public Health This article is licensed under the Creative Commons Attribution-NonCommercial 4.0 International License (CC BY-NC) ( https://karger.com/Services/OpenAccessLicense ). Usage and distribution for commercial purposes requires written permission. PMC Copyright notice PMCID: PMC13241156  PMID: 42254541 Abstract Introduction The application of AI in public health surveillance presents a transformative potential for more efficient work methodologies, offering new capabilities for the detection, notification, and response to public health threats. However, it is important to be aware of the associated risks and the ethical and legal challenges linked to AIs use in such a sensitive area as public health. The general objective of this study was to explore the application of AI in public health surveillance in Portugal, as viewed by Portuguese experts. Methods The methodological approach employed was a qualitative study, utilising a descriptive and exploratory investigation. A content analysis was performed on 28 anonymised semi-structured interviews. This process identified concepts, themes, key ideas, and emergent patterns. The findings were then grouped into categories and subcategories, which allowed us to highlight significant consensuses and meaningful insights. Results Experts recognised AIs transformative potential in enhancing predictive capacity, automating data processes, and supporting decision-making. However, they highlighted critical limitations of the Portuguese system, notably entrenched reactivity, fragmented infrastructures, under-utilisation of primary care and personal device data, and insufficient surveillance of non-communicable diseases. Key barriers include poor data quality and interoperability, absence of comprehensive data governance, shortage of AI-skilled professionals, resistance to organisational change, and limited financial sustainability. Ethical concerns, privacy, algorithmic bias, and explainability were emphasised as central to AIs legitimacy. Comparisons with international experiences revealed that progress depends less on technological readiness than on systemic reform, data harmonisation, and strong governance aligned with European Union (EU) and World Health Organization (WHO) frameworks. Conclusion AI can shift Portuguese public health surveillance from reactive to predictive, but only if supported by robust data ecosystems, clear governance, and investment in human capital. A national roadmap is required, prioritising interoperability, sustainable financing, and ethical implementation, to ensure that AI serves as a complement to, rather than a replacement for, human judgement in protecting population health. Keywords: Artificial intelligence, Digital transformation, Public health surveillance, Decision making in public health, Portugal Plain Language Summary The application of AI tools in public health surveillance presents a transformative potential for more efficient work methodologies, offering new capabilities for the detection, notification, and response to public health threats. The aim of this study was to explore the view of experts about the application of AI tools in public health surveillance. 28 experts were interviewed and shared their views. Addressing Portugal’s current public health surveillance system, they described its problems such as outdated information systems and issues with data quality and access. Experts believe AI could significantly improve how public health is monitored and acted upon. However, they also raised important concerns about data security and privacy, potential bias in AI systems, and the need for transparency in how AI operates. This highlights the importance of new legislation and ethical guidelines, like those proposed in the AI Act, to guide AI use. Key challenges include managing data effectively, ensuring system interoperability and training healthcare professionals. For AI to be used successfully, it requires human oversight, robust data systems, specific legislation, investment in new technologies and collaboration between several organisations. In conclusion, while AI can offer considerable benefits to public health surveillance, it is constrained by current limitations in data and information systems. Portugal needs a digital transformation in health, making sure that new technology is introduced while also supporting the people involved. Training professionals and having strong ethical and legal regulations are crucial for using AI responsibly and effectively in public health in Portugal. Introduction Public health confronts increasingly complex challenges driven by globalisation, climate change, rising chronic diseases, and the rapid spread of infections and antimicrobial resistance [ 1 , 2 ]. Effectively addressing these issues requires new policies, robust structural preparedness, enhanced inter-institutional coordination, and the creation of targeted emergency programmes [ 3 ]. In this scenario, Artificial Intelligence (AI) emerges as a promising technology for public health, particularly within surveillance systems [ 4 , 5 ]. Public health surveillance is defined as the systematic collection, analysis, and interpretation of health-related data needed to plan, implement, and evaluate public health practice. Given AIs potential to transform this process, surveillance is one of the essential public health operations poised to benefit most from its application. AI enables the early detection of outbreaks and the prediction of epidemiological trends, facilitates more efficient resource management, and allows for rapid and targeted responses [ 6 , 7 ]. It can also identify patterns and anomalies within large datasets due to its capacity to process and organise big data. This includes unstructured information via natural language processing, thus reducing manual work and improving analytical accuracy. The resulting intelligence can impact the development of evidence-based health policies, strengthening health systems’ response and decision-making capabilities, and promoting public health [ 8 , 9 ]. Finally, AIs automation of data collection and processing can facilitate the fulfilment of international reporting objectives, notably to the European Centre for Disease Prevention and Control (ECDC) [ 10 ]. In Portugal, the Directorate-General for Health and the National Institute of Health Doutor Ricardo Jorge have been proactive in implementing AI in public health. Notable projects include AutoCod for automated death coding [ 11 – 13 ], the use of AI for event-based surveillance by the Public Health Emergency Centre [ 14 ], and the REVIVE project for monitoring disease-transmitting vectors [ 15 ]. Moreover, Portugal participates in European Union (EU) initiatives like EU-HIP to strengthen information system interoperability with the ATHINA platform of HERA (European Health Emergency Preparedness and Response Authority), which aims to increase effectiveness in public health emergencies [ 16 ]. Other examples in Portugal include CLIKODE for clinical coding in hospital settings [ 17 ] and the integration of Large Language Model tools (e.g., ChatGPT, Copilot, Gemini) across various public services [ 18 ]. Despite its transformative potential, AI implementation faces significant ethical and systemic challenges, particularly concerning the processing of sensitive data. AIs effectiveness is contingent upon data quality; incomplete or inaccurate data risks algorithmic bias, which can lead to incorrect conclusions and amplify existing inequalities [ 19 ]. The sensitivity of health data raises significant privacy and security concerns as the risk of re-identification remains even after anonymisation, necessitating strict sharing and usage protocols [ 20 ]. Furthermore, many AI models are complex and difficult to interpret (the “black box” phenomenon), meaning a lack of explainability compromises decision-making transparency and fosters distrust [ 21 ]. AI adoption requires robust technological infrastructures and a workforce skilled in both AI and public health [ 22 ]. AIs rapid evolution also demands a clear regulatory framework to assign responsibility for errors (whether attributed to the technological system, medical professionals, or institutional bodies) [ 23 ]. Finally, while trust and acceptance are crucial, over-reliance on AI could diminish human judgement, which remains irreplaceable in public health contexts [ 24 ]. While the World Health Organization (WHO) recognises AIs “enormous potential” in health systems, it cautions against its misuse and the risk of exacerbating inequity [ 6 , 25 ]. The handling of sensitive data requires rigorous risk mitigation aligned with national and international legislation. Core ethical principles, including beneficence, non-maleficence, human autonomy, and justice, are enshrined in the Treaty on European Union (EU) [ 26 ]. In Portugal, AI use is regulated to ensure respect for fundamental rights, balancing principles such as explainability, safety, transparency, and accountability [ 27 ]. Processes must actively avoid bias and discrimination, and decisions with significant impact should be reported, appealable, and auditable. Furthermore, health data management must strictly comply with the General Data Protection Regulation (GDPR) and national legislation [ 28 , 29 ]. Global health organisations such as the Pan American Health Organization (PAHO) and WHO stress that AI in public health must adhere to evidence standards, with transparency and external evaluation to minimise bias [ 30 ]. The principles for AI use in public health include being people-centred, ethical, transparent, with data protection, scientific integrity, open, shareable, non-discriminatory, and human-controlled [ 31 ]. While various AI subfields (e.g., machine learning, large language models, and computer vision) benefit public health, their successful implementation requires efficient and interoperable information systems [ 32 ]. The necessity of robust regulatory frameworks, sustained investment, and strategic partnerships remains paramount [ 33 , 34 ]. Crucially, AI applied to public health surveillance – particularly for outbreak prediction, resource allocation, or population monitoring – may be classified as a High-Risk System under the EU AI Act [ 35 ]. This classification underscores the critical importance of analysing expert perceptions regarding specific implementation requirements within the reality of Portuguese public health. The UN Educational, Scientific and Cultural Organization (UNESCO) highlights the necessity of ensuring that AI systems serve people, society, and the environment. Preventing harm and promoting human rights and the Sustainable Development Goals is a priority [ 36 ]. The debate surrounding AIs role in public health is, therefore, vital to accelerate its safe and effective implementation in surveillance [ 37 , 38 ]. Within Portugal, ongoing structural reforms, such as the implementation of Local Health Units [ 39 ], coupled with challenges of population mobility and oscillating disease patterns, require accurate public decision-making and optimisation of resources [ 9 , 40 ]. The use of innovative and safe AI in surveillance is essential to modernise procedures, promote effective response, and increase the resilience and sustainability of the National Health Service (NHS) [ 20 , 41 , 42 ]. This study is framed by the digital transition scenario, driven by the Portuguese Recovery and Resilience Plan, and it underscores the technological and innovative component of public health surveillance systems [ 43 ]. It aims to reflect on the challenges of digitalisation so that Portugal can fully leverage AI in building a more resilient health system with surveillance preparedness [ 44 , 45 ]. The findings are intended to contribute to the development of efficient, sustainable, and safe AI-based methodologies, and to the establishment of partnerships aligned with international strategies and the goals of digital transformation in health [ 46 , 47 ]. Consequently, the general objective of this study was to explore the perspectives of Portuguese experts on the application of AI in public health surveillance in Portugal, with the following specific objectives. To describe the current scenario of public health surveillance in Portugal and information and data systems within this context, identifying its needs and challenges. To explore the perceptions and experiences of experts on AIs use as a tool to improve the performance of the public health surveillance system and the achievement of its objectives, identifying an ideal AI-driven public health surveillance system in Portugal. To gather insights on AIs contributions to public health decision-making and recommendations for policymakers on applying AI in public health surveillance in Portugal. Materials and Methods Study Design The methodological approach employed was a qualitative study following the Consolidated Criteria for Reporting Qualitative Research (COREQ) and the Standards for Reporting Qualitative Research (SRQR), utilising a descriptive and exploratory investigation [ 48 ]. Data were collected through semi-structured interviews with 28 intentionally selected Portuguese experts. Population and Sample The target population comprised Portuguese experts with relevant experience and knowledge in public health surveillance and/or health information systems in Portugal, specifically at the intersection of AI, data science, information technology, public health, health, medicine, academia, and health management. Professionals with proven experience in AI in health, public health physicians, or related fields and specialities representing local, regional, and central levels were represented. Other professionals with expertise in the public health sector, including nurses and pharmacists, biomedical and hospital administrators, engineers, and other data science and computing specialists with relevant experience in information systems and digital health, including representatives from academia, the public, and private sectors, were also represented ( Table 1 ). Table 1 provides a grouped characterisation of the study participants, illustrating the multidisciplinary and hierarchical diversity of the experts interviewed across public health, information technology, and administrative fields. Table 1. Grouped characterisation of the 28 study participants (source: authors) Main training category Major area of expertise and professional experience N Medicine Public health/epidemiological surveillance/hospital management and administration/digitalization/data science and AI 14 Medicine Other specialties/university/management/health information technology and AI 3 Nursing University/hospital management and administration/digitalization/data science and AI 4 Pharmacy Healthcare consulting and digitalization/health information technology and AI 2 Management/administration Hospital administration/health consulting 2 Engineering/information technology/computer science Health data analytics and AI/university/project management 3 Total ​ 28 Open in a new tab Experts with specific knowledge about the Portuguese context and who demonstrated availability, interest, and consent to participate were included. An intentional, convenience, and snowball sampling strategy was used to include key experts, prioritising the depth of knowledge obtained over statistical generalisation. The recruitment of participants aimed to represent services, hierarchical levels, and areas of specialisation and expertise in a balanced and diversified manner to gather the necessary insights for the study’s objectives, along with a geographical distribution to ensure representativeness from the north, centre, and south regions of the country. Sample size was determined by verifying sample saturation, which occurred when the same ideas and concepts became recurrent in the answers obtained ( n = 28). Within the variability of the experts' areas of expertise, one strand focused on public health surveillance ( n = 14) was combined with specialised perspectives on data science, digital tools, and AI in surveillance and public health ( n = 14). Data Collection For data collection, a script for the semi-structured interviews was prepared, aiming to address the study’s objectives. Questions were formulated according to key themes for a broad exploration of the study topic, based on the literature review and identified research gaps. The questions sought to collect, explore, and characterise the perception, experiences, and perspectives of experts and specialists on the application of AI in public health surveillance. This ranged from describing the current state of public health surveillance in Portugal and the challenges and determinants of AI application in this context, to their future perspectives on the ideal AI-driven public health surveillance system, its impact on public health decision-making, and finally, recommendations for policymakers (online suppl. Additional file 1; for all online suppl. material, see https://doi.org/10.1159/000550578 ). The validity of the interview script was ensured by a pilot test with a key expert, which helped better align the questions with the study’s objectives and impacted the objectivity of the final questionnaire; this first interview was not included in the analysis. Once interviewees were selected, prior contact was made with an invitation sent, which supported the informed consent of the interviewees. This guaranteed a free, informed, and voluntary decision to participate, under conditions of privacy and individuality. The interviewee was responsible for choosing the interview time and format (in person or by video-conference with recording and full transcription), both consented to through the signing of informed consent. The semi-structured interviews followed the script and were conducted in Portuguese by the principal investigator, in an empathetic and non-directive manner to encourage the exploration of themes. Data Analysis Researcher reflexivity and rigour: The multidisciplinary background of the authors (public health, hospital administration, information technology, and data science) was acknowledged, and reflexivity was considered throughout the study to minimise bias in data interpretation, ensuring that the findings were grounded in the empirical data while being critically contrasted with the established literature and technological feasibility. At the outset of the study, the authors shared an inherent enthusiasm for the technological potential of AI to enhance public health functions, tempered by a cautious awareness of the structural limitations within the Portuguese context. The average duration of the interviews was 40 min. The interviews were audio-recorded, transcribed verbatim, and anonymised (E1-E28). A qualitative content analysis was carried out with the exploratory purpose of generating new knowledge. The analysis stages began with (1) reading the interviews, (2) coding, (3) identifying and categorising themes, (4) defining patterns and ideas, and (5) preparing the report. In the first phase, after reading, key ideas and concepts were identified, grouped by themes and sub-themes. Their meaning and relevance were validated through discussion based on national, European, and international bibliographic research, with a focus on regulation and relevant scientific publications for result validation. To better organise the report and structure the themes, data structuring began with the variables that formed the basis for interview construction, allowing for broad and generic initial coding, followed by focused and specific coding, leading to a more detailed analysis of expert perspectives. The identification and description of themes and successive patterns until sample saturation ( n = 22) and theme exhaustiveness ( n = 28) allowed for the identification of emerging consensus patterns, as well as possible divergences and nuances in expert views. In turn, the densification of themes is reflected by the greater or lesser development of each theme and sub-theme. In a final phase of report construction, excerpts from the interviews derived from coding were translated into English to enrich the results chapter. Rigor and Qualitative Credibility The entire methodology of implementation, development, and analysis of the study aimed for a systematic and coherent approach to ensure methodological rigour and results endowed with credibility, transferability, and reliability. The results were presented with objectivity and structure, framing them within the study objectives and the literature review on AI in public health surveillance, ensuring a contextualised, enriched, and comparative understanding of the object of study. The structuring of the results was based on a description of the current Portuguese scenario, the collection of insights into the potential of AI applications, and finally, future perspectives and recommendations for policymakers, to frame the theme and give visibility to the potentials and challenges of AI in the Portuguese context of public health surveillance. Results The 28 semi-structured interviews were conducted with a diverse group of Portuguese experts, comprising two main cohorts: 14 specialists from Public Health and other health disciplines, and 14 experts in hospital management, administration, information science, data, and health computing. This structure ensured a comprehensive range of insights. To prevent individual identification, predominant professional backgrounds were grouped into broad categories based on their relevance to AI in public health surveillance ( Table 1 ). Table 1 provides a grouped characterisation of the study participants, illustrating the multidisciplinary and hierarchical diversity of the experts interviewed across public health, IT, and administrative fields. Experts with relevant experience across multiple areas were categorised by their primary or most pertinent contribution to the study’s context. Qualitative content analysis identified four main themes that were developed separately: “Current state of public health surveillance in Portugal,” “Potential of AI application,” “Limitations and challenges for AI implementation,” and “Future perspectives and recommendations to policymakers.” Consensus emerged regarding critical limitations of current information systems. Experts highly valued AIs transformative potential, particularly its capacity for predictive analysis. Data quality and accessibility were identified as the unanimous critical barriers to implementation. Finally, data quality and interoperability were prioritised as crucial future implementation factors. The identified themes and sub-themes are summarised in the conceptual mind map ( Fig. 1 ). Figure 1 summarises the qualitative content analysis, presenting a comprehensive mind map of the four main themes and their sub-themes, which structured the experts' perspectives on AI application in Portuguese surveillance. Fig. 1. Open in a new tab Mind map illustrating the themes and sub-themes identified regarding the application of AI in public health surveillance in Portugal as viewed by experts (source: authors). Current State of Public Health Surveillance in Portugal The vast majority of respondents characterised the current state of public health surveillance in Portugal as fundamentally “reactive and fragmented,” “focused on past events,” and “lacking a predictive perspective” (E15, E3). Supporting information systems were seen as numerous but plagued by a generalised “lack of interoperability,” resulting in “unequal and incomplete,” “dispersed and unharmonised” data (E12, E17). Daily problems included the “inefficiency of manual records” and “bureaucracy” (E14). The system’s weakness was attributed to a “lack of strategy” rather than a lack of funding (E20). Furthermore, public health work was often perceived as requiring “greater dissemination” and visibility (E21). Outdated and Ineffective Information Systems Most interviewees voiced criticism of the existing systems, which were described as “outdated” (E5, E20, E24, E26, E28), “fragmented” (E5, E9, E22, E28), and “siloed”(E6, E10, E11, E16, E23, E24), with a pervasive “lack of interoperability” (E2, E7, E12, E13, E22, E23, E24, E26, E27, E28) and “poor data governance” (E22). These isolated systems fail to communicate, leading to concerns about the poor use of the “absurd amount of information” contained in “unstructured data” (E28). Much data collection remains “manual” (E16, E20), often through “forms and surveys” (E10, E26), causing “duplication of records” (E10, E26), “delays in disease notification” (E4, E10), and resulting in a “distributed data pool” whose “collection and processing is still very manual and painstaking” (E20). The current state is characterised by the underutilisation of existing tools, where the “heavy burden and manual reporting” prevent information from “reaching the end-user” in the field, keeping it “locked in higher hierarchical levels” and hindering “timely decision-making based on as much information as possible” (E16). There is a “massive underuse” of existing data (E16, E26, E28), including information from “primary healthcare and personal devices” (E26), and a “lack of communication across care levels” (E9). Surveillance is considered “unsophisticated,” based on “old concepts” and achieving “low efficiency” (E27, E28), and “practically zero use” within the context of the “heterogeneous database” (E28). However, the National Epidemiological Surveillance System (SINAVE) was identified as a strength, considered “well framed” for communicable disease surveillance, being accessible to “all notifying doctors and all laboratories” (E2). Despite this robustness, a “growing challenge in the surveillance of chronic non-communicable diseases” (E2, E8), such as obesity and sedentary lifestyle, is underlined, where the data are “less robust” and there is difficulty in “understanding their prevalence” (E2). Although SINAVE forms the basis for disease notification, the information is not sufficiently integrated for “clinical case management” (E17). Reactive and Non-Proactive Surveillance The system was consistently described as reactive, focussing on “counting what has already happened,” monitoring “past events,” and “recording deaths” and “emergency response” (E15, E24, E27, E2). This creates a significant “gap in the ability to predict and anticipate outbreaks or trends” (E15, E25, E28). Experts expressed a strong desire for proactivity and the generation of “information for action” (E24, E27). Reporting processes remain predominantly “manual and bureaucratic” (E11), leading to poor “data and information quality” (E10, E11, E20). Data access is generally difficult, “often taking months to obtain” (E20). Limited Human and Operational Capacity A “lack of specialised human resources” and “excessive administrative burden” were underlined as factors “hindering the implementation of more advanced approaches” (E14, E16). The system “burdens professionals with administrative tasks” in data entry (E27), preventing them from having time to analyse data (E16). This capacity shortage is compounded by a persistent “resistance to change” among professionals (E11, E17, E27), who are “often reluctant to fully use existing platform functionalities” (E17). The “lack of human resources” was ultimately identified as the “main problem of the NHS” (E9). Potential of AI in Public Health Surveillance Experts shared an optimistic outlook and a broad consensus on AIs potential to revolutionise surveillance, transforming it from reactive to proactive, making it “more active, more dynamic, and real-time” (E7). AI was viewed as being “here to stay” (E7). However, national public health surveillance currently shows an almost complete absence of AI projects, with only “isolated attempts” and “examples in the clinic” (E27, E28). Predictive and Proactive Analytics Consensus was overwhelming regarding AIs crucial role in predictive analytics. This includes “outbreak prediction” (E1, E2, E3, E8, E10, E12, E21, E28), “pattern identification” (E10), “developing risk models” (E25), performing “projections” (E27), “trend detection” (E26, E28), and enabling “real-time monitoring” and “early warnings” (E3, E19; E12, E27, E28). Broadly, AI is also seen as vital for patient safety and quality, managing risks such as “hospital infections,” reducing “adverse events” (E10, E11), predicting “demand peaks in emergency services” (E14, E15), and detecting “risky behaviours” (E2). By facilitating “targeted and personalised interventions,” AI makes health more “predictive and preventive” (E2, E12). The technology is perceived as a facilitator that “helps the public health doctor to be faster and more efficient” without replacing human judgement (E4, E10, E12). Process Optimisation and Efficiency AI was recognised for its ability to “automate data collection and analysis” (E6, E16, E18), “reduce administrative burden” (E16), and “optimise resource planning and allocation” (E1, E8, E11, E12). Experts highlighted its “speed” and “agility” (E27, E28) to process vast data volumes (E1, E27), “identify patterns,” and deal with “multidimensionality” in ways impossible for humans (E27). AI can also “streamline processes” to “free up time for decision-making” and plays a crucial role via Natural Language Processing (NLP) for “unstructured data” (E1, E2, E11, E27, E28). This capacity is fundamental for tackling complex “chronic diseases.” allowing the “detection of non-obvious patterns” (E1, E2, E8, E19, E28). The potential to “optimise clinical processes” (E10) and “resource allocation” (E1, E8, E28) was also mentioned. Ultimately, AI is seen as a tool to “optimise and streamline,” though its effectiveness is entirely dependent on “data quality and integration” (E12). Supporting Human Decision-Making and Personalisation AI can provide “clinical and public health decision support” (E8, E21, E24), “freeing up time” and enabling “evidence-based decisions” and “personalised public health” (E11, E28). Key benefits include improving “speed in decision-making” (E3) and offering data-driven insights for assertive, individualised interventions (E15). Experts also noted its potential for using chatbots for health promotion (E1, E28) and for “detecting health inequalities” and “assessing health policy impact” (E1). AI serves as a “decision support” tool that provides “suggestions for decisions or paths to follow” (E27, E1, E10, E21). Crucially, experts stressed that the “final decision will remain human” (E21, E27), and AI must never replace the human decision-maker (E4, E8, E10, E2). AI is viewed as a “complement” that “frees up professional time for the patient” (E7, E22), by optimising routine tasks. Experts consider it to reduce burnout, and to “facilitate the integration of the currently fragmented system” by “connecting primary care and hospitals” (E8, E9). The scope of AI support includes “clinical decision support,” “resource optimisation” (e.g., operating room management), “emergency triage,” “demographic data analysis,” and “risk management,” ultimately culminating in the ability to “personalise health” (E15). Value Creation and Sustainability AI was viewed as a means of “creating value” (E25), “strengthening the capacity and resilience of the health system” (E3), and promoting “sustainability and proactivity” (E25). AI contributes significantly to “resource optimisation” (E3, E5, E8, E10). By “predicting outbreaks” and “identifying inefficiencies in processes” (E4, E8, E13), AI facilitates faster, “less expensive action” (E4, E8). Beyond epidemiological surveillance, its usefulness extends to “planning team allocation,” “managing infrastructures.” and “reducing waste” (E5). This allows for targeted interventions and better results (E1). The identified potentialities, however, confront practical limitations in applying AI to public health data and decisions ( Fig. 2 ). Figure 2 offers a schematic summary, visually condensing the current reactive state, the transformative potential, and the critical systemic challenges identified by experts as fundamental to AI implementation in Portugal. Fig. 2. Open in a new tab Schematic summary of the current state, potential, challenges, and limitations of the application of AI in public health surveillance in Portugal (source: authors). Limitations and Challenges of AI in Public Health Surveillance Despite the general optimism, respondents identified a number of complex challenges, particularly systemic and cultural barriers, that currently hinder the widespread adoption of AI in public health. Experts underlined the “incipience” of AI application in Portuguese public health compared to countries like the USA and China, which “are advancing very quickly in this new revolution” (E1). Barriers include “lack of motivation and bureaucracy within the public service for the development of AI solutions,” stressing the need for rewarding developers and adopting a “bottom-up” approach involving end-users (E23). Data Quality and Accessibility This was the most frequently mentioned and critical barrier (E10, E20, E22). The lack of “quality data”, “longitudinal and population data,” and “interoperability” between systems constitutes a fundamental impediment (E25, E2, E7, E9, E13, E17, E22, E26, E27, E28). Data resides in fragmented “silos” (E10, E16, E24), aggravated by a prevailing “culture of non-data sharing” (E24). There is a need to create “very good and very well-processed” data structures (E27). Experts noted that the “biggest challenge” involves the “quality and integration of basic data,” including “data cleaning and standardisation” (E13, E28). The complexity of managing “data from many sources” and the “difficulty of building interoperable systems” were also highlighted (E22). Core concerns revolve around the “risk of data bias,” the “dependence on the quality of input data”, and the risk of the “machine overrides the person” (E3). The community dimension of adoption was noted, requiring the population to be “aware” and “want to share” data within a necessary “data ecosystem” (E5, E6). This fragmentation is compounded by the “perception that the data belongs to each one” (E6). The dichotomy between the “registered” (formal) and the “implemented” (real) is also noted, reinforcing the idea that data quality is not limited to its completeness but also its integrity and veracity (E10) for safe AI use. Finally, a prior failure in a public health Information System due to a lack of maintenance serves as a warning about the necessity of long-term sustainability (E12). Governance, Ethics and Regulation There is widespread concern about the security and “privacy of data” (E4, E5, E6, E7, E10, E11, E15, E26, E27, E28) and the risk of algorithmic bias that can “reproduce inequalities” (E5, E11). This highlights the need for social justice principles and a “democratisation of AI” built on fair, representative data (E11). Many argue for the urgent need for specific “health data legislation”, including “chapters dedicated to AI and public health” (E26). The success of AI relies on society's ability to debate and establish a “robust regulatory and ethical framework that guides its responsible development and use” (E11). Experts highlighted the necessity for “transparency and accountability in algorithms” (E11, E15), warning against the “black box” phenomenon (E6, E11, E15, E27, 28), potential “bias in algorithms” and AIs potential for “disinformation” (E28). The need for an “ethical discussion” before implementation is fundamental (E5, E11). Public administration faces “very lengthy and laborious processes” in software procurement, and the “European regulatory framework” is perceived as “very tight,” potentially “stifling innovation” (E1). Human Resources and Training A consensus emerged on the critical challenge of the “lack of trained people” to “use” AI (E8, E27), specifically “professionals with specific skills” in AI within the health sector (E21, E22, E27, E28). Simultaneously, experts identified “resistance to change” (E11, E17, E27), “cultural resistance and inertia of the system” (E28), and a “fear that AI will replace jobs” (E22). Sensitising professionals to AIs value is crucial to promoting its “acceptance” (E19, E7). The “training and capacity building of HR” to “develop” and “use” AI must be continuous and practical (E5, E7, E28, E25). The concern is that AI should “improve the autonomy” of the professional, not replace them (E5). There is also a need for “training professionals to fill in data correctly” (E10) and for the establishment of “complete teams” to promote “brainstorming” and “task allocation to explore new approaches” (E14). A “massive investment in digital literacy” is crucial so that both professionals and the population understand AIs benefits and risks, including the ability to “know how to ask the right questions of AI” (E15). Fundamentally incorporating the “experience and critical judgement of human beings” into the decision-making process (E15). The lack of “internal support” and the absence of “innovation, data recording, and information management sections” within institutions are significant barriers (E16). Investment and Financing While some experts believe that “there is no shortage of money in Portugal; there is a shortage of strategy” (E20), “significant and continuous investment” in infrastructure and AI use is crucial (E25). Significant barriers to implementation include the “high costs” and “financial investment” (E28) associated with “buying software” or, more complexly, “developing it,” which can be “quite costly” (E1). “Budget and investment” were cited as a priority, given the current context of “insufficient funding” and “lack of political will” (E27, E12). Crucially, the absence of a system that “rewards” professionals who “have good ideas” leads to a “lack of motivation” in the public service, resulting in inefficient “top-to-bottom” solutions without the “involvement of users” (E23). Organisational Mindset and Culture An “old mentality” (E26) and the absence of a “competitive logic” within the NHS (E24) are perceived as obstacles to innovation. “Bureaucracy,” “delays in decisions,” and a lack of “strategic vision” (E9, E20) are identified as impeding factors. The “cultural resistance and inertia of the system” (encapsulated by the phrase “it has always been like this”) and the lack of “strategy” and “political will” (E28, E27, E28) are central points. The recognition of generational resistance and the need for “training and capacity building” (E5) highlights the cultural and change management dimension, which is considered critical as the technology itself. The importance of “giving professionals a voice in systems development” is emphasised as key for technology acceptance (E9). Finally, the “absence of monitoring and evaluation of implemented measures” is classified as “waste,” as it prevents learning and continuous improvement (E23). Future Prospects: The Ideal AI-Powered Surveillance System The interviewees' vision is unanimous: a public health surveillance system driven by AI that is “predictive, proactive, integrated, and automated” (E12). This system would be “capable of cross-referencing dispersed data and anticipating scenarios and needs” (E6). AI is viewed as having the “ability to internalise learning” (E6, E16) and to transform data into “knowledge for action” (E6). The ideal model would be “highly interoperable and integrated, where AI operates on quality data,” and where innovation is driven by a “bottom-up” approach (E23). The implementation should be a collaborative process, “built by all stakeholders” (E26). For the development of this ideal system, experts offer key recommendations to policymakers ( Fig. 3 ). Figure 3 provides a schematic representation, highlighting the sequential and interdependent nature of the core recommendations necessary for policymakers to safely and effectively integrate AI into Portuguese public health surveillance. Fig. 3. Open in a new tab Schematic representation of key recommendations for Portuguese policymakers (source: authors). Ideal Surveillance System with AI Experts suggested an “integrated” system (E13, E17, E24, E26) – a “data ecosystem” (E6) offering “real-time information” (E3, E10, E19). This system would enable “predictive analysis” (E10, E15, E25) and a “360° view” of individual and wider population (E6, E13, E16, E26, E28), capable of “anticipating” and responding “on time”, with “continuous monitoring capability” (E6, E16, E17, E20). They envision a “much more agile and faster” system for “immediate action” underpinned by “quality data” (E10), “much more complete data” from “all sources” (E27, E28), featuring “centralised information” and “interoperability” (E2, E13, E17). This would be “integrated into a single national platform” (E2, E28), incorporating automation, “real-time alerts and forecasts” (E12, E28), “personalised public health” (E28), and a multidisciplinary approach “involving all stakeholders” in its construction (E26). The future is seen as “active and automated with AI” (E8, E13), prioritising “prevention” (E8, E13) and ultimately aiming to “reduce disease and mortality” (E8). The ideal system is “centralised, interoperable, and with AI integrated at all levels,” with a “reinforced focus on prevention” (E11). For instance, AI could be utilised to “improve case records and follow up”, “identify patients at risk for screening for chronic diseases”, and “optimise the management of patient flows in emergency rooms, including the analysis of waiting times” (E14). Crucial Factors for Implementation Data quality and interoperability, legislation and governance, investment and implementation, capacity building and mindset change, and leadership and strategy were cited as critical factors. Data Quality and Interoperability . Top priority lies in ensuring robust “data governance” (E6, E16, E22), guaranteeing “data quality” (E10, E20), and fostering the “harmonisation” and “interoperability” of systems (E7, E11, E13, E22, E26). The quality of input data and governance will ultimately determine AIs success (E6, E10). Legislation and Governance . The creation of “ a health data law with a chapter dedicated to artificial intelligence and a chapter dedicated to public health” (E26) is viewed as the indispensable legal foundation. The necessity for “transparency” and “ethics” (E11, E15) in AI use is vital, requiring “leadership and governance” (E5, E11) and “multidisciplinary collaboration” (E5). The need for “safeguards that are necessary from an ethical and regulatory point of view” is also stressed (E25). Investment and Implementation . Experts advised to “start using” AI, even if imperfectly initially, and to “invest in that use” to enable “continuous improvement” and the “identification of flaws” (E25). It is crucial to avoid “reinventing the wheel” (E8) and ensure implementation occurs “as soon as possible, but sustainably” (E5), driven by a “link between engineering and medicine” and “pilot projects” (E21, E24). This must lead to a “redefinition of basic health processes” (E26), fundamentally reorganising practice. Empowerment and Mindset Change . While “great curiosity” has been noted, it is acknowledged that “there are professional groups that may feel in the crosshairs of technologies” (E1). It is essential to invest in reinforcing and “training human resources” (E9, E14, E22), both by promoting the perceived value of AI and demystifying concerns over job displacement. There is an imperative to recruit and “pay well” (E21) personnel with “different training, cross-training, [and] poly-crossing,” particularly in the areas of “data, technology, and computing” (E16, E21). Furthermore, “the voice of the professional” (E21) must be heard during the process of “changing the mindset of many people” (E21). Leadership and Strategy . The need for “political will” (E2, E12, E14) and a robust “strategic vision and leadership” (E9, E20) to drive innovation is paramount. This must involve “team engagement and management” to achieve “health gains” (E18). AI must function as a human-guided “instrument,” wherein the system is expected to “adapt and correct the trajectory,” ensuring new technologies are not a “source of new problems” (E6). Crucially, a key factor mentioned was “the ability to bring people together” and motivate them to “create solutions, fostering a sense of ownership” (E23). Key Recommendations for Policymakers AI is highlighted as a “strategic opportunity to strengthen the capacity and resilience of our NHS” (E3). Implementation should be predicated on robust “data governance” and the principle of “knowledge to action,” alongside clearly defined “process dimensions” (E6, E22) to ensure the “continuous improvement” of public health (E6). The importance of “giving the power of decision and development to creators” (E23) already within the public service, enabling them to develop their own “solutions from scratch” rather than resorting to “external consultancies for off-the-shelf solutions” (E23) was also emphasised. Prioritise Data Quality and Governance Investment in the “base of the pyramid, data quality” was deemed an imperative step, requiring a “long-term commitment” of time and money (E10). A cultural shift is needed to value the “quality of information in the daily practice of health professionals” (E10). Investment in the “ability to aggregate data, to conserve data and to work with it” is crucial, an area where Portugal is still “a little below what is expected” (E2). It is crucial that the “base is created” for any predictive model to be effective, necessitating prior investment in “data harmonisation” and “interoperability,” especially in “longitudinal and population databases” (E7, E17). Furthermore, the “security and control” of AI and the “trust of professionals and the population” are fundamental (E4), which underscores the suggested creation of a “health data law” (E26). Implement Sustainably To effectively “implement artificial intelligence as soon as possible, but with sustainability” (E5), “technological tools should be implemented at the national level from the outset”. This is because creating “technologies of this type at scale always offers an advantage” by allowing “economies of scale” and ensuring the software is “common to all Local Health Units,” thereby promoting “interoperability with other applications” and “data comparability” (E1). Any advance will crucially depend on the “political will” that translates into concrete actions (E2, E9) as AI has the potential to “reduce the burden on health spending for the Portuguese Government” (E2). The term “sustainability” refers not only to financial terms but also to the ability to maintain AI with “adequate human and material resources” in the long term (E5). Experts recommended to “implement AI on the condition that it guarantees the necessary internal support and know-how,” allowing “crossing between teams and management of technology implementation teams,” reinforcing areas like “data, technology, and computing,” to avoid exclusive dependence on “external contracts” (E16). Investment in Implementation Experts advised to “start using [AI] and invest in its ongoing use” (E25) to foster continuous learning and improvement. Policymakers should “clearly back AI,” but to “begin with small steps” that offer “high quality” and utilise “well-processed data,” specifically targeting “epidemiological surveillance and outbreak prediction” (E27). It was also suggested to commence with “small projects to demonstrate value” and that “public health chatbots” be recommended from official entities (E28). The “integration of data” and investment in AI was strongly recommended as a measure that “saves lives” and “optimises management,” stressing the urgency given the current system's ineffectiveness (E13). One striking comment highlighted that “there's no shortage of money in Portugal; there's a shortage of strategy,” underscoring the vital need for a “culture of rigorous monitoring and evaluation of health policies” (E20). Collaboration and Competitiveness Bureaucracy is seen as a major obstacle to “public health reforms,” as despite well-designed plans and numerous initiatives, in practice, “nothing is ever implemented” (E8). Conversely, developing solutions from scratch instead of “taking advantage of what already exists” leads to an “unnecessary delay” (E8). Promoting collaboration and competitiveness are pivotal factors: “developing AI in Europe and trying to bring European companies into collaboration” with health systems and introducing a “competitive logic” within the NHS (E24) could significantly boost successful implementation. Focus on the Training of Professionals A central concern is Portugal’s lack of preparedness for the “logarithmic” or “exponential” learning curve of AI that could “exceed the country’s capacity to train professionals” to “work with and absorb it” (E8). The focus must be on the “continuous involvement of professionals” (E12) in “training health professionals to work with AI,” ensuring they don’t feel “replaced” (E22). Additionally, “training in team and process management, as well as in law […] would enable them to better understand the system and actively contribute to its optimisation” (E9). It was recommended that, “AI, although a powerful tool, will only be effective if accompanied by a substantial investment in human resources and in forming multidisciplinary and complete teams capable of operationalising and integrating it meaningfully into the day-to-day of public healthcare” (E14), as technology alone will not suffice to solve pressing operational challenges. Training for AI in public health involves “raising people’s awareness” by “using these tools and showing what they’re for”, specifically through “presenting two or three examples of how these tools are used and what the outputs were, and how this can support decision-making” (E19). It is necessary to “convey to policymakers the importance” of training personnel so they “don't feel that these types of tools take away some of their jobs, but rather make them work with them as a complement to their daily tasks” (E22). Discussion While Portuguese experts acknowledge AIs transformative potential in public health surveillance, their reflections critically translate theoretical opportunities and barriers into the practical realities of implementation. The compiled insights reveal a complex landscape of opportunities and challenges in Portugal, tempered by ethical, regulatory, and operational constraints within the national system. Ultimately, the value of AI depends on its ethical and transparent integration. This imperative must not be overlooked in the pursuit of innovation, a concern aligned with global discussions [ 49 – 52 ]. Current State of Surveillance in Portugal: Entrenched Reactivity and Data Fragmentation The most critical finding of this study, consistently corroborated by recent analyses of Portugal’s digital health architecture [ 42 , 53 – 57 ], is the characterisation of national public health surveillance as fundamentally reactive and ineffective. This limitation is inseparable from the fragmentation and obsolescence of its information systems. Experts described a scenario where systems operate in silos, severely restricting interoperability and the continuous flow of data [ 53 ]. This represents not only operational inefficiency but also a strategic barrier to AI implementation. The system prioritises retrospective event-counting over anticipatory action. Furthermore, its reliance on manual and bureaucratic data collection produces duplication and under-reporting, undermining both data quality and timeliness [ 54 ]. Thus, Portugal’s primary challenge lies not in the absence of advanced algorithms, but in the fragility of its data foundations. Without reliable, high-quality, and complete data, AI applications remain inherently vulnerable [ 58 , 59 ]. These operational shortcomings impose a heavy administrative burden on public health professionals, diverting scarce time and expertise from field investigation and preventive action [ 44 , 60 , 61 ]. In Portugal, SINAVE is the most widely used system in public health surveillance for notifiable diseases. Despite its evolution through digitisation, it remains insufficient to address contemporary needs [ 56 , 57 ]. This reinforces the perception that digitalisation has largely replicated pre-existing manual processes, failing to achieve the structural transformation required for modern, data-driven public health. A critical strategic failure emerges in the entrenched “culture of non-sharing” and the marked under-utilisation of existing datasets, particularly longitudinal monitoring and integration with primary care datasets [ 11 , 55 , 62 ]. Specifically, information streams from primary care and personal devices are consistently neglected. These streams could substantially enhance the sensitivity and representativeness of surveillance efforts if appropriately governed and interoperable [ 42 , 55 , 59 , 63 , 64 ]. This absence of rich, longitudinal data contrasts sharply with AI potential to deliver the predictive insights desired by practitioners [ 60 ]. Experts also highlighted a persistent imbalance between communicable disease surveillance and the neglected monitoring of non-communicable diseases. The latter remains under-resourced and under-instrumented. This strategic gap undermines public health priorities and ignores the global shift in disease burden [ 1 , 65 ]. Transformative Potential of AI: From Theory to Practical Necessity Portuguese experts broadly align with the international consensus that AI offers transformative technical capabilities for public health surveillance. These include high-throughput data processing, pattern recognition across heterogeneous sources, natural language processing of clinical free text, and computer-vision-enabled analysis of remote sensors [ 66 – 69 ]. These capabilities are viewed as catalysts for change, enabling predictive analytics, early warning systems, and automated triage to accelerate the detection and response to emerging threats [ 70 – 72 ]. Automation can also significantly reduce administrative burdens [ 73 ], echoing OECD perspectives on building resilience and sustainability of health systems through anticipatory capacity [ 71 , 74 ]. International examples, such as AI-supported epidemic intelligence and automated syndromic surveillance, reinforce the role of algorithmic tools. These tools can supplement human expertise with timelier signals, decision support, and the promotion of personalised health through accurate risk stratification, enabling more assertive and individualised interventions [ 75 , 76 , 77 ]. However, this transformative potential must be anchored in realism. Experts adopt a cautious stance [ 78 ], stressing that technical promise alone is insufficient. AI cannot compensate for missing longitudinal, population-level, interoperable data [ 58 , 59 , 79 ]. The consistent warning from Portuguese experts is clear: without foundational reform of data and governance, AI risks remaining a collection of isolated pilots rather than a driver of systemic transformation. Algorithms, however sophisticated, are only as effective as their underlying data. This underscores the triple imperative: data, governance, and human capital. Data Foundation and Interoperability: From Reactivity to Predictive Capacity Consensus among experts is unequivocal: moving surveillance from a reactive model to a proactive, predictive paradigm requires more than model development. It demands longitudinal, high-quality, interoperable data, along with institutional mechanisms to translate model outputs into action [ 58 , 59 , 79 ]. Data harmonisation and interoperability are the most fundamental prerequisites for reliable AI in public health [ 80 , 81 ]. Experts and literature converge on the principle that governance and data quality must precede technological deployment [ 59 ]. Critically, AI efforts often falter when applied to incomplete, biased, or siloed datasets, problems that cannot be solved by algorithmic sophistication alone [ 58 , 59 ]. Predictive accuracy and public health utility of AI depend on representative training data, continuous model validation, and robust pipelines for updating and recalibration [ 66 , 71 , 82 ]. This need is compounded by profound governance risks. The absence of comprehensive health data legislation [ 46 , 83 , 84 ] and a persistent non-sharing culture [ 62 ] demand urgent regulatory action. Data security, algorithmic accountability, and transparency [ 85 – 87 ] are not theoretical concerns but legal and ethical imperatives. Therefore, experts call for a robust framework of data harmonisation and governance aligned with the EU AI Act and WHO guidelines, as prerequisites for reliable AI in surveillance [ 6 , 59 , 71 ]. Policy measures like the European Health Data Space offer pathways to enhance data sharing and reuse. However, this requires matching interoperability standards and incentives for data custodians to participate [ 46 , 83 ]. Governance, Ethics and Regulation: Aligning Trust, Law, and Practice Experts emphasised that the ethical, legal, and social implications of AI, such as privacy, bias, explainability, and accountability, are not abstract issues but central determinants to AI deployment [ 7 , 19 , 25 , 31 , 52 ]. Portugal faces the dual challenge of aligning national practice with General Data Protection Regulation obligations [ 28 ] while preparing for the obligations and opportunities introduced by the EU AI Act [ 35 , 80 ]. Portuguese respondents framed the AI Act as a timely lever to standardise governance but one that must be operationalised with national specificity to build professional and public trust [ 20 , 31 , 36 ]. Operationalising governance requires concrete mechanisms, including transparent documentation, audit trails, algorithmic impact assessments, and accountability protocols for automated decisions, alongside accessible channels for public redress [ 21 , 22 , 86 , 87 ]. Workforce, Capacity, and Evaluation: The Human Capital Imperative The shortage of professionals with expertise in data science, AI, and digital epidemiology emerged as a critical bottleneck [ 3 , 61 , 88 ]. Experts call for structured training, career pathways for technical professionals within public health, and collaborative models pairing domain specialists with data scientists [ 74 , 88 , 89 ]. According to experts, AI interventions should meet the same evidentiary standards as other public health tools. This includes prospective validation, external review, and monitoring for unintended effects [ 82 , 90 ]. Such evaluation frameworks safeguard against bias and ensure accountability and transparency, thereby reinforcing professional and public trust [ 91 – 93 ]. Experts recommended embedding human oversight across the model lifecycle, ensuring that accountability for decisions rests with professionals rather than algorithms. In line with WHO guidance and calls for explainable AI in health, this ethical non-negotiable underscores that AI should complement, never replace, clinical and public health judgement [ 21 , 31 , 94 ]. Financing and Multisectoral Partnerships: Pragmatic Pathways to Scale Long-term sustainability was identified as a decisive barrier in the public health context, characterised by budgetary limitations. AI integration requires continuous investment in infrastructure, model maintenance, and workforce development; these costs exceed short-term innovation budgets [ 74 , 95 ]. Portuguese experts recommended blended financing, combining public funding, EU mechanisms (such as the Portuguese Recovery and Resilience Plan [ 43 ]), and governed public–private partnerships [ 96 ]. Regulatory sandboxes and pilot networks were repeatedly cited as pragmatic solutions. These enable safe testing while generating evidence for broader scaling [ 19 , 97 , 98 ]. These approaches balance innovation with accountability, bridging the gap between pilot projects and systemic integration. Institutional Culture and Organisational Change: The Global AI Landscape A recurrent and distinctive theme was the weight of institutional culture in constraining innovation. Experts emphasised resistance to change, the absence of a national AI strategy in health, and limited inter-institutional collaboration as decisive barriers [ 14 , 15 , 20 , 44 , 51 ]. Comparisons with international experiences highlight Portugal’s relative disadvantage. Nordic countries, for example, rely on long-established health registries and data-sharing norms to support scalable AI [ 65 ]. Germany has similarly prioritised interoperability as a pillar of digital reform [ 99 ]. Other jurisdictions, such as Singapore, Canada, and Australia, have advanced rapidly by coupling governance frameworks with sustained investment and cross-sectoral partnerships [ 100 – 102 ]. In contrast, Portugal lacks a long-term funding strategy and a coherent national roadmap [ 44 , 95 ]. Meanwhile, the Food and Drug Administration (FDA) in the USA has promoted regulatory sandboxes for AI as a safe innovation mechanism [ 103 ]. These examples confirm Portugal’s challenges are structural and cultural, not merely technological. Policy Implications: A Strategic Roadmap A broad consensus was reached regarding the ideal system: it should be integrated and real-time, offering a predictive and 360° view of the population's health. It must also be multidisciplinary and participatory, involving professionals from different areas and levels of activity, from the local, regional, and central entities [ 64 , 104 ]. The future of AI-powered public health surveillance in Portugal requires a strategic and collaborative approach, in a scenario where systemic and cultural challenges outweigh technological ones ( Table 2 ). Table 2 details the resulting strategic roadmap, outlining pragmatic, staged policy interventions across data, governance, workforce, and financing to transition surveillance from a reactive to a predictive model. Table 2. Roadmap: a guide regarding the application of AI in public health surveillance in Portugal (source: authors) Roadmap: AI in public health surveillance in Portugal Public health surveillance: current state AI: transformative potential AI: challenges and limitations Future and recommendations Ineffective information systems Archaic In silos Lack of interoperability Manual data collection Duplication of records Notification delays Data underutilisation Reactive surveillance Focus on the past Emergency response Lack of prediction Operational and human limitations Lack of specialised human resources Excessive administrative burden Predictive analytics Outbreak prediction Demand peaks Risk models Population screening Early detection Real-time monitoring Optimisation and efficiency Automation: data collection/analysis Reducing administrative burden Optimised resource allocation Big data processing Personalised decision support Clinical and public health decision support Personalised, assertive interventions Public health chatbots Value and sustainability Strengthening system resilience Proactive and sustainable system Data: quality and access Lack of quality data Absence of longitudinal, Population-based data Non-interoperable Non-sharing culture Governance and ethics Privacy and security Lack of health data legislation AI responsibility Algorithmic transparency Human resources training Shortage of AI skills Resistance to change Need for training Financing Ongoing investment Lack of long-term support Organisational culture Traditional mindset Lack of competitiveness Lack of strategic vision Ideal system with AI Integrated Real-time Predictive 360° vision Multidisciplinary Participatory Success factors Data quality and interoperability Strong legislation and governance Early adoption with iteration Training and mindset change Strategic leadership Recommendations to decision-makers Prioritise data and governance Implement sustainably Promote collaboration and competitiveness Empowering healthcare professionals Open in a new tab Drawing on international best practice and expert testimony, a staged national roadmap emerges: (1) immediate investments in data harmonisation, interoperability, and minimum datasets for priority conditions; (2) parallel development of governance instruments ensuring a robust ethical and regulatory framework for AI (algorithmic accountability and alignment with EU law); (3) concomitant workforce development programmes and the establishment of applied data science units within public health agencies; (4) phased deployment of interoperable pilot systems with rigorous evaluation criteria; and (5) scale-up alongside sustainable financing mechanisms and continuous citizen engagement, ensuring transparency and legitimacy. Each stage requires clear metrics of success and independent evaluation to guard against premature over-deployment and to ensure public health benefit [ 31 , 59 , 90 , 80 ]. General Recommendation for Portugal Portuguese experts recognise AIs potential to transform public health surveillance. However, they insist that technical innovation must be embedded within a reformed data ecosystem, robust governance, and deliberate organisational change. The country’s immediate priorities are clear: data harmonisation, EU/WHO-aligned governance, workforce capacity-building, and sustainable financing. If secured, AI can evolve from promising pilots to durable public health gains, enhancing timeliness, prediction, and equity in surveillance while preserving human oversight and accountability. This study offers a timely reflection for public health and technology development professionals, hospital administrators, policymakers, and regulators in Portugal. It also serves as a warning regarding the exponential pace of AI adoption internationally, urging Portugal to accelerate its paradigm shift. Promoting AIs safe and ethical implementation can strengthen surveillance systems and align Portugal with global trajectories [ 105 – 110 ]. Limitations As an exploratory qualitative study, the findings represent the detailed perspectives of a selected group of national experts. Consequently, the results possess transferability to similar contexts or systems, but they are not statistically generalisable to the entire public health or IT workforce. The focus on high-level strategic challenges meant that technical implementation details were not explored in depth. Furthermore, the insights gathered reflect the views and opinions held at the time of the interviews and may evolve as technology and regulations progress. A crucial limitation inherent to qualitative research is the potential influence of the research team's background and preconceptions (researcher reflexivity). The multidisciplinary expertise of the authors was acknowledged, and reflexivity was maintained throughout the process of data collection and analysis to minimise bias in interpretation. Nevertheless, the potential influence of researcher reflexivity remains an intrinsic limitation of the study’s design. The results are specific to the Portuguese context, although they might be transferable to other countries with similar characteristics, especially those aligned with EU frameworks. Conclusions This qualitative study confirms both the promise and the fragility of AI adoption in Portuguese public health surveillance. Experts broadly agree that AI could enhance prediction, detection, and response, but the reality of outdated, fragmented systems and entrenched reactivity casts doubt on the country’s readiness for large-scale implementation. The analysis suggests that AI, while often portrayed as a technological solution, risks becoming another layer of complexity if structural weaknesses are not first addressed. Findings point to a central paradox: although Portugal recognises AIs transformative potential, its current surveillance infrastructure and governance frameworks remain unready to translate this potential into practice. The lack of data harmonisation, inadequate legislation, and a persistent culture of non-sharing undermine the very foundations upon which AI depends. Without decisive reform, technology is likely to remain confined to pilot projects with limited systemic impact. The implications are significant. For Portugal, progress depends less on technological readiness than on political will, organisational reform, and sustained investment in human capital. AI should not be seen as a shortcut to modernisation but as a tool whose effectiveness is contingent upon governance, ethics, and the strengthening of core data systems. Unless these priorities are urgently addressed, AI risks reinforcing existing inequities and inefficiencies rather than resolving them. The challenge ahead is, therefore, not merely technical but strategic: to ensure that AI supports a shift from reactive to predictive, equitable, and accountable surveillance, with human oversight at its core. Building on these findings, future research should focus on several priority areas to accelerate safe and effective AI adoption in Portugal: investigating sustainable financing models for AI in public health; evaluating the impact of training and digital literacy initiatives among health professionals; developing detailed legal and ethical frameworks for AI in Portuguese public health; formulating strategies to change the organisational culture for effective AI implementation; optimising interoperability and health data quality; conducting comparative studies with other EU member states where AI is already applied to public health surveillance, providing crucial insights into transferable practices and context-specific adaptations. Acknowledgement The authors thank all the experts and specialists who participated in the study. Statement of Ethics This study protocol was reviewed and approved by the Ethics Committee of NOVA-ENSP, Lisbon, Portugal, Approval No. ID:09/2025. The study was based on original data. The authors confirm that all methods were performed in accordance with the relevant guidelines and regulations and confirm that written informed consent was obtained from participants to participate in the study. Conflict of Interest Statement The authors have no conflicts of interest to declare. Funding Sources This study was not supported by any sponsor or funder. Author Contributions Liliana Certo conducted data collection and composed the manuscript. Maurício Alves and Teresa Magalhães edited and performed a critical review of the manuscript and provided guidance. All authors have read and approved the final version of the manuscript. Funding Statement This study was not supported by any sponsor or funder. Data Availability Statement Data are available to the investigators upon request and approval by the principal investigator of the original study. Semi-Structured Interview Script (Annex 1) is available in Additional File 1. Supplementary Material. Supplementary_material-Suppl.1-s1.docx (32.1KB, docx) References 1. Institute for Health Metrics and Evaluation. Global Burden of Disease Collaborative Network . Global Burden of Disease Study 2019 (GBD 2019): results. Seattle, WA: IHME; 2020. [ Google Scholar ] 2. World Health Organization . Public health preparedness and response: a framework for resilient health systems. Geneva: WHO; 2022. [ Google Scholar ] 3. European Centre for Disease Prevention and Control . ECDC assessment of public health workforce capacity in prevention and control of infectious diseases in the EU/EEA: report on 2024 survey. Stockholm: ECDC; 2025. [ Google Scholar ] 4. Duarte R, Lopes F, Alves F, Aguiar A, Monteiro H, Pinto M. COVID-19 in Portugal: the strategy. Braga: UMinho Editora; 2022. [ Google Scholar ] 5. Vitorino G, Cordeiro J, Magalhães T. Digital transformation in its various dimensions. In: Magalhães T, editor. Digital transformation in healthcare. Coimbra: Edições Almedina; 2021. [ Google Scholar ] 6. Jiang F, Jiang Y, Zhi H, Dong Y, Li H, Ma S, et al. Artificial intelligence in healthcare: past, present and future. Stroke Vasc Neurol. 2017;2(4):230–43. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 7. Dankwa-Mullan I. Health equity and ethical considerations in the use of artificial Intelligence in public health and medicine. Prev Chronic Dis. 2024;21:240245. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 8. World Health Organization . Big data and Artificial intelligence to achieve global health coverage: international consultation about ethics: meeting report, 12–13 October 2017, Miami, Florida, USA: WHO; 2018. [ Google Scholar ] 9. European Commission . Communication from the Commission to the European Parliament, the Council, the European Economic and Social Committee and the Committee of the Regions: White Paper on Artificial Intelligence: a European approach to excellence and trust. Brussels: European Commission; 2020. [ Google Scholar ] 10. European Centre for Disease Prevention and Control . Annual epidemiological reports (AERs) [Internet]. Stockholm: ECDC; 2025. [cited 2025 Jun 4]. Available from: https://www.ecdc.europa.eu/en/publications-data/monitoring/all-annual-epidemiological-reports [ Google Scholar ] 11. Portugal. Ministry of Health . AUTOCOD Project: automating the coding of causes of death. Lisbon: Ministry of Health; 2023. [ Google Scholar ] 12. Duarte F, Martins B, Pinto CS, Silva MJ. Deep neural models for ICD-10 coding of death certificates and autopsy reports in free-text. J Biomed Inform. 2018;80:64–77. [ DOI ] [ PubMed ] [ Google Scholar ] 13. Pita-Ferreira P, Godinho-Simões D, Pinto-de-Carvalho C, Duarte F, Fernandes E, Casaca- Carvalho P, et al. Real-time classification of causes of death using AI: sensitivity analysis. JMIR AI. 2023;2:e40965. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 14. Portugal. Directorate-General for Health. Contingency Plan for Public Health Responses. Lisbon: Directorate-General for Health; 2022. [ Google Scholar ] 15. Portugal. Ministry of Health. National Institute of Health Doutor Ricardo Jorge . REVIVE Project: identification of disease vectors [Internet]. Lisbon: INSA; 2013. [cited 2025 Jun 4]. Available from: https://www.insa.min-saude.pt/projectos/revive/ [ Google Scholar ] 16. European health emergency preparedness and response authority (HERA). HERA's IT platform (ATHINA) [Internet]. Brussels: HERA; 2025 [cited 2025 Jun 4]. Available from: https://health.ec.europa.eu/latest-updates/hera-it-platform-athina_en 17. University Hospital of São João . CLIKODE: automatic clinical coding processing [Internet]. Porto: CHUSJ; 2024. [cited 2025 Jun 4]. Available from: https://www.chusj.pt/projectos/clikode/ [ Google Scholar ] 18. Microsoft. Copilot in healthcare [Internet]. Redmond, WA: Microsoft; 2025. Available from: https://www.microsoft.com/en-us/ai/copilot-healthcare [ Google Scholar ] 19. World Health Organization . Artificial Intelligence for Health: supporting countries to deploy responsible AI technologies to accelerate equitable health for all. Geneva: WHO; 2024. [ Google Scholar ] 20. Portugal. Ministry of Health . White Paper Artificial Intelligence in Health in Portugal: regulation, impacts and future perspectives. Lisbon: SPMS; 2025. [ Google Scholar ] 21. Sadeghi Z, Alizadehsani R, CIFCI MA, Kausar S, Rehman R, Mahanta P, et al. A review of explainable artificial intelligence in healthcare. Comput Electr Eng. 2024;118(A):109370. [ Google Scholar ] 22. van Leersum CM, Maathuis C. Human-centred explainable AI decision-making in healthcare. J Respons Technol. 2025;21:100108. [ Google Scholar ] 23. Chen Y, Esmaeilzadeh P. Generative AI in medical practice: in-depth exploration of privacy and security challenges. J Med Internet Res. 2024;26:e53008. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 24. Office of the National Coordinator for Health Information Technology . Health IT playbook: artificial intelligence. Washington, DC: ONC; 2023. [ Google Scholar ] 25. World Health Organization . WHO urges caution and ethics in AI use for health [Internet]. Geneva: WHO; 2023. [cited 2025 Jun 4]. Available from: https://www.who.int/news/item/16-05-2023-who-calls-for-safe-and-ethical-ai-for-health [ Google Scholar ] 26. Treaty on European Union . Art. 2. Off J Eur Union. 2012;C 326:13–390. [ Google Scholar ] 27. Portugal. Government of Portugal. National Strategy for Artificial Intelligence. Lisbon: Government of Portugal; 2020. [ Google Scholar ] 28. Regulation (EU). 2016/679 of the European parliament and of the Council of 27 April 2016 on the protection of natural persons with regard to the processing of personal data and on the free movement of such data, and repealing Directive 95/46/EC (General Data Protection Regulation). Off J Eur Union. 2016. L 119. http://data.europa.eu/eli/reg/2016/679/2016-05-04 29. European Group on Ethics in Science and New Technologies. Statement on Artificial Intelligence and Human Values. Luxembourg: Publications Office of the European Union; 2018. [ Google Scholar ] 30. Pan American Health Organization . Artificial Intelligence in health: opportunities and challenges. Washington, DC: PAHO; 2021. [ Google Scholar ] 31. World Health Organization . Ethics and governance of artificial intelligence for health: WHO guidance. Geneva: WHO; 2021. [ Google Scholar ] 32. Pan American Health Organization . Q&A on artificial intelligence for supporting public health: reference tool to support the exchange of information and promote open conversations and debates. Washington, DC: PAHO; 2024. [ Google Scholar ] 33. Olawade DB, Wada OJ, David-Olawade AC, Kunonga E, Abaire O, Ling J. Using Artificial Intelligence to improve public health: a narrative review. Front Public Health. 2023;11:1196397. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 34. Panteli D, Adib K, Buttigieg S, Goiana-da-Silva F, Ladewig K, Azzopardi-Muscat N, et al. Artificial Intelligence in public health: promises, challenges, and an agenda for policymakers and public health institutions. Lancet Public Health. 2025;10(5):E428–E432. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 35. European Union . Regulation (EU) 2024/1689 of the European Parliament and of the council on a European approach for Artificial Intelligence (AI Act). Off J Eur Union. JO L. 2024/1689. 36. UNESCO . Recommendation on the ethics of artificial intelligence. Paris: UNESCO; 2021. [ Google Scholar ] 37. European Commission . Communication from the Commission to the European Parliament, the Council, the European Economic and Social Committee and the Committee of the Regions: the Digital Compass 2030: europe’s digital decade. Brussels: European Commission; 2021. [ Google Scholar ] 38. Pham T. Ethical and legal considerations in healthcare AI: innovation and policy for safe and fair use. R Soc Open Sci. 2025;12(5):241873. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 39. Decreto-Lei n° 102/2023 . Diário da República. Diário da República n.° 215/2023. I. Série; 2023-11-07. p. 4–20. https://data.dre.pt/eli/dec-lei/102/2023/11/07/p/dre/pt/html . [ Google Scholar ] 40. Alves M, Seringa J, Silvestre T, Magalhães T. Use of Artificial Intelligence tools in supporting decision-making in hospital management. BMC Health Serv Res. 2024;24(1):1282. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 41. United Nations . The 2030 Agenda for sustainable development. New York, NY: United Nations; 2015. [ Google Scholar ] 42. Estrela M, Ferreira PL, Roque F, Herdeiro MT. “ Simplification, decentralization, proximity ” : a critical analysis of the digital health framework in Portugal through expert interviews. Int J Med Inform. 2025;201:105962. [ DOI ] [ PubMed ] [ Google Scholar ] 43. Portugal. Recovery and resilience plan. Lisbon: Portuguese Republic; 2021. [ Google Scholar ] 44. Portugal. Directorate-General for Health. National Health Strategy 2020-2030. Lisbon: Directorate-General for Health; 2021. [ Google Scholar ] 45. Portuguese Association of Hospital Administrators . Strategic agenda for the future of precision medicine in Portugal: guidelines for implementation 2020-2023. Lisbon: APAH; 2019. [ Google Scholar ] 46. European Commission . Communication from the Commission to the European Parliament, the Council, the European Economic and Social Committee and the Committee of the Regions: a European strategy for data. Brussels: European Commission; 2020. [ Google Scholar ] 47. World Health Organization . Global strategy on digital health 2020-2025. Geneva: World Health Organization; 2021. [ Google Scholar ] 48. Vilelas J. Investigação: o processo de construção do conhecimento. 3rd ed. Lisboa: Sílabo; 2020. [ Google Scholar ] 49. Seringa J, Hirata A, Pedro AR, Santana R, Magalhães T. Health care professionals and data scientists’ perspectives on a machine learning system to anticipate and manage the risk of decompensation from patients with heart failure: qualitative interview study. J Med Internet Res. 2025;27:e54990. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 50. Floridi L, Cowls J, Beltrametti MA, Chatila R, Chazerand P, MacIntosh B, et al. AI4People: ethical guidelines for a sustainable AI. AI Soc. 2019;34(1):15–26. [ Google Scholar ] 51. Portugal. Directorate-General for Health. National Health Plan 2030 Sustainable Health: from all for all. Lisbon: Directorate-General for Health; 2022. [ Google Scholar ] 52. Harishbhai Tilala M, Kumar Chenchala P, Choppadandi A, Kaur J, Naguri S, Saoji R, et al. Ethical considerations in the use of Artificial Intelligence and machine learning in health care: a comprehensive review. Cureus. 2024;16(6):e62443. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 53. Morris G, Snider D, Katz M. Integrating public health information and surveillance systems. J Public Health Manag Pract. 1996;2(4):24–7. [ DOI ] [ PubMed ] [ Google Scholar ] 54. Madans JH, Hunter EL. Improving and integrating data systems for public health surveillance. J Public Health Manag Pract. 1996;2(4):42–4. [ DOI ] [ PubMed ] [ Google Scholar ] 55. European Data Protection Board . Guidelines 07/2020 on the concepts of controller and processor in the GDPR. Brussels: EDPB; 2020. [ Google Scholar ] 56. Pinto-Leite P. Public health information systems and SINAVE. Lisbon: Directorate-General for Health; 2024. [ Google Scholar ] 57. European Centre for Disease Prevention and Control . AI in infectious disease surveillance: opportunities and challenges. Stockholm: ECDC; 2022. [ Google Scholar ] 58. Demirol D, Das R, Hanbay D. A key review on security and privacy of big data: issues, challenges, and future research directions. SIViP. 2022;17(4):1335–43. [ Google Scholar ] 59. Costa T, Borges-Tiago T, Martins F, Tiago F. System interoperability and data linkage in the era of health information management: a bibliometric analysis. Health Inf Manag. 2025;54(3):214–26. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 60. Jungwirth D, Haluza D. Artificial Intelligence and public health: an exploratory study. Int J Environ Res Public Health. 2023;20(5):4541. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 61. World Health Organization . AI for epidemic intelligence: a landscape review. Geneva: World Health Organization; 2023. [ Google Scholar ] 62. Ferreira A, Antunes L, Chadwick D, Correia R. Grounding information security in healthcare. Int J Med Inform. 2010;79(4):268–83. [ DOI ] [ PubMed ] [ Google Scholar ] 63. Torres DR, Wermelinger ED, Ferreira AP. Application of Artificial Intelligence in primary health care: scoping review and critical analysis. Saude debate. 2025;49(145):e10070. [ Google Scholar ] 64. Yousefi F, Dehnavieh R, Laberge M, Gagnon MP, Ghaemi MM, Nadali M, et al. Opportunities, challenges, and requirements for artificial Intelligence (AI) implementation in Primary Health Care (PHC): a systematic review. BMC Prim Care. 2025;26(1):196. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 65. Sundgren S, Reponen J. Nordic eHealth benchmarking report 2022. Copenhagen: Nordic Council of Ministers; 2022. [ Google Scholar ] 66. Badawy M, Ramadan N, Hefny HA. Predictive analytics in healthcare using machine learning and deep learning techniques: a research. J Electr Syst Inf Technol. 2023;10(1):40. [ Google Scholar ] 67. Portela D, Brito DV, Monteiro H. Using digital technologies in response to the COVID-19 pandemic in Portugal. Port J Public Health. 2022;39(3):170–4. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 68. Pan American Health Organization . Artificial Intelligence in public health. Washington, DC: PAHO; 2022. [ Google Scholar ] 69. World Medical Association . WMA declaration on ethical considerations regarding health databases. Ferney-Voltaire: WMA; 2023. [ Google Scholar ] 70. Sharma P, Maheshwari M, Chaturvedi B. Artificial intelligence in healthcare: a paradigm shift towards personalised treatment. Int Adv Res Sci Commun Technol. 2024;4(8):124–31. [ Google Scholar ] 71. MacIntyre CR, Chen X, Kunasekaran M, Quigley A, Lim S, Stone H, et al. Artificial intelligence in public health: the potential of epidemic early warning systems. J Int Med Res. 2023;51(3):3000605231159335. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 72. Jia P, Liu S, Yang S. Innovations in public health surveillance for emerging infections. Annu Rev Public Health. 2023;44:55–74. [ DOI ] [ PubMed ] [ Google Scholar ] 73. Wang Q, Su M, Zhang M, Li R. Integrating digital technologies and public health to fight COVID-19 pandemic: key technologies, applications, challenges and outlook of digital healthcare. Int J Environ Res Public Health. 2021;18(11):6053. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 74. OECD . Artificial Intelligence in health: key issues, challenges and opportunities. Paris: OECD Publishing; 2023. [ Google Scholar ] 75. Sharma A, Adekunle BI, Ogeawuchi JC, Abayomi AA, Onifade O. AI-driven patient risk stratification models in public health: improving preventive care outcomes through predictive analytics. Int J Multidiscip Res Growth Eval. 2023;4(3):1123–30. [ Google Scholar ] 76. Cordeiro JV. Artificial intelligence and precision public health: a balancing act of scientific accuracy, social responsibility, and community engagement. Port J Public Health. 2024;42(1):1–5. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 77. Pereira-Morales AJ, Rojas LH. Risk stratification using Artificial Intelligence: could it be useful to reduce the burden of chronic kidney disease in low- and middle-income Countries? Front Public Health. 2022;10:999512. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 78. Nunes R, Ricou M, Godinho I, Neves M. Artificial intelligence (AI): social concerns, ethical proposals and political orientations: white paper. Lisbon: National Council of Ethics for the Life Sciences (CNECV); 2024. [ Google Scholar ] 79. Li L, Back E, Lee S, Shipley R, Mapitse N, Elbe S, et al. Balancing risks and opportunities: data-Empowered-health ecosystems. J Med Internet Res. 2025;27:e57237. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 80. European Parliament and Council . Regulation (EU) 2024/1689 of 13 June 2024 laying down harmonised rules on artificial intelligence and amending Regulations (EC) no 300/2008, (EU) no 167/2013, (EU) no 168/2013, (EU) 2018/858, (EU) 2018/1139 and (EU) 2019/2144 and Directives 2014/90/EU, (EU) 2016/797 and (EU) 2020/1828 (Artificial Intelligence Act). Off J Eur Union. L 2024/1689, Jul 12, 2024. Available from. https://eur-lex.europa.eu . [ Google Scholar ] 81. European Centre for Disease Prevention and Control . Digital transformation in public health surveillance. Stockholm: ECDC; 2023. [ Google Scholar ] 82. National Academies of Sciences, Engineering, and Medicine . AI in Healthcare: the winding path to trust. Washington, DC: The National Academies Press; 2023. [ Google Scholar ] 83. European Health and Digital Executive Agency (HaDEA) . Health data space. Brussels: HaDEA; 2024. [ Google Scholar ] 84. Pinto C, Martins J, Martins H. From big data to smart health: putting data to work for the public’s health-data strategy for next generation Portuguese national health service. Lisbon: SPMS; 2019. [ Google Scholar ] 85. Canelas-Pais M, Almeida JC, Araújo SM, Rafael FM, Cruz-Correia R, Rodrigues PP. Artificial intelligence in healthcare. In: Pais A, Vitorino C, Nunes S, Cova T, editors. Artificial Intelligence for drug product lifecycle applications. New York, NY: Academic Press; 2025. p. 235–71. [ Google Scholar ] 86. European Union Agency for Cybersecurity . Deploying pseudonymisation techniques [Internet]. Athens: ENISA; 2022. [cited 2025 Jun 4]. Available from: https://www.enisa.europa.eu/publications/deploying-pseudonymisation-techniques [ Google Scholar ] 87. European Union Agency for Cybersecurity . AI in healthcare: risks, opportunities and recommendations. Brussels: ENISA; 2022. [ Google Scholar ] 88. Khan Rony MK, Akter K, Nesa L, Islam MT, Johra FT, Akter F, et al. Healthcare workers' knowledge and attitudes regarding artificial intelligence adoption in healthcare: a cross-sectional study. Heliyon. 2024;10(23):e40775. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 89. European Centre for Disease Prevention and Control . Core competencies in applied infectious disease epidemiology in Europe. Stockholm: ECDC; 2022. [ Google Scholar ] 90. European Centre for Disease Prevention and Control . Artificial Intelligence for public health: a scoping review. Stockholm: ECDC; 2022. [ Google Scholar ] 91. Groseclose SL, Buckeridge DL. Public health surveillance systems: recent advances in their use and evaluation. Annu Rev Public Health. 2017;38:57–79. [ DOI ] [ PubMed ] [ Google Scholar ] 92. Ramezani M, Takian A, Bakhtiari A, Rabiee HR, Ghazanfari S, Mostafavi H. The application of artificial intelligence in health policy: a scoping review. BMC Health Serv Res. 2023;23(1):1416. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 93. Hoelscher SH, Pugh ANURSES. Embracing artificial intelligence: a guide to artificial intelligence literacy for the nursing profession. Nurs Outlook. 2025;73(4):102466. [ DOI ] [ PubMed ] [ Google Scholar ] 94. World Health Organization . Strengthening population health surveillance: a tool for selecting indicators to signal and monitor the wider effects of the COVID-19 pandemic. Copenhagen: WHO Regional Office for Europe; 2021. [ Google Scholar ] 95. United Nations Development Programme . Digital transformation strategy 2022-2025. New York, NY: UNDP; 2022. [ Google Scholar ] 96. Haltaufderheide J, Ranisch R. The ethics of ChatGPT in medicine and healthcare: a systematic review on Large Language Models (LLMs). npj Digit Med. 2024;7(1):183. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 97. European Centre for Disease Prevention and Control . Amended ECDC Strategy 2021–2027. Stockholm: ECDC; 2023. [ Google Scholar ] 98. European Commission . Communication from the Commission to the European Parliament, the Council, the European Economic and Social Committee and the Committee of the Regions: Shaping Europe’s digital future. Brussels: European Commission; 2020. [ Google Scholar ] 99. Germany. Federal Ministry of Health . Digital health strategy in Germany: strengthening interoperability and data use. Berlin: Bundesministerium für Gesundheit (BMG); 2023. [ Google Scholar ] 100. Singapore. Ministry of Health . National Artificial Intelligence Strategy: health sector applications. Singapore: Government of Singapore; 2023. [ Google Scholar ] 101. Canadian Institute for Health Information . Harnessing AI for health surveillance in Canada. Ottawa: CIHI; 2023. [ Google Scholar ] 102. Australian Digital Health Agency . Artificial Intelligence in health: national roadmap. Canberra: ADHA; 2024. [ Google Scholar ] 103. U.S. Food and Drug Administration . Artificial Intelligence and machine learning in health care: a regulatory sandbox approach. Silver Spring, MD: FDA; 2023. [ Google Scholar ] 104. Lopes MA, Martins H, Correia T. Artificial intelligence and the future in health policy, planning and management. Int J Health Plann Manage. 2024;39(1):3–8. [ DOI ] [ PubMed ] [ Google Scholar ] 105. Felix-Cardoso J, Vasconcelos H, Pereira Rodrigues P, Cruz-Correia R. Excess mortality during COVID-19 in five European countries and a critique of mortality data analysis. medRxiv. 2020. [Preprint]. 106. European Commission . Digital education action plan 2021-2027. Brussels: European Commission; 2020. [ Google Scholar ] 107. Kumar R. Transforming public health: shifting from reactive to proactive. J Pub Health Nutri. 2022;5(01):105. [ Google Scholar ] 108. Jerfy A, Selden O, Balkrishnan R. The growing impact of natural language processing in healthcare and public health. Inquiry. 2024;61:469580241290095. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 109. OECD . Health at a glance 2023: OECD Indicators. Paris: OECD Publishing; 2023. [ Google Scholar ] 110. Silva G. Artificial intelligence for the surveillance of non-communicable chronic diseases [Thesis]. São Paulo: Faculty of Public Health; 2023. [ Google Scholar ] 111. United Nations . The future of AI in healthcare. New York, NY: UN; 2023. [ Google Scholar ] Associated Data This section collects any data citations, data availability statements, or supplementary materials included in this article. Supplementary Materials Supplementary_material-Suppl.1-s1.docx (32.1KB, docx) Data Availability Statement Data are available to the investigators upon request and approval by the principal investigator of the original study. Semi-Structured Interview Script (Annex 1) is available in Additional File 1. Articles from Portuguese Journal of Public Health are provided here courtesy of Karger Publishers ACTIONS View on publisher site PDF (2.9 MB) Cite Collections Permalink RESOURCES Similar articles Cited by other articles Links to NCBI Databases 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

Related documents

• Epidemiology and Public Health Intelligence#10933
• AI-BASED EARLY DETECTION AND SURVEILLANCE OF INFECTIOUS DISEASES IN PAKISTAN: A PUBLIC HEALTH DATA ANALYTICS APPROACH#20799
• Between the hype and harm: does artificial intelligence in health offer solace or further exclusion for marginalised populations in Sub-Saharan Africa? A scoping review#651914
• Between the hype and harm: does artificial intelligence in health offer solace or further exclusion for marginalised populations in Sub-Saharan Africa? A scoping review.#428259
• Harnessing Artificial Intelligence in Health Research in Low-Income and Middle-Income Countries: Potential and Caution.#335035