Machine-Learning Assessment of the Predictive Value of Inflammatory Biomarkers for Cognitive Impairment in an Older Hispanic Adult Cohort

Small clinical tabular datasets require interpretable machine learning because deep learning is often impractical and ensemble models can be difficult to inspect. A key pitfall is that statistical significance does not necessarily imply predictive utility. Using data from the Panama Aging Research I