Trustworthy AI in Digital Health: A Comprehensive Review of Robustness and Explainability

This review examines trustworthy AI for digital health, with emphasis on robustness and explainability across the AI lifecycle. It situates these topics alongside fairness, accountability, and privacy, and discusses application-specific concerns in intensive care, neonatal health, and metabolic health. The paper surveys methods for robustness under data scarcity and distribution shifts, plus explainability techniques including feature attribution, gradient-based interpretation, and counterfactual explanations. It also discusses trustworthy AI in the era of LLMs and evaluation concepts such as validity, fidelity, and diversity.