EXPLAINABLE AI FOR THERAPEUTIC DECISION-MAKING AND PRESCRIPTION SAFETY: A LONGITUDINAL FRAMEWORK FOR CLINICAL DECISION SUPPORT

Background: Clinical artificial intelligence has been evaluated mainly through diagnosis, triage, image interpretation, and isolated question answering. Therapeutic decision-making has a different structure: it converts clinical reasoning into action through drug choice, dose, route, timing, contraindication screening, monitoring, reassessment, escalation, de-escalation, and discontinuation. An explainable system that names a diagnosis but does not account for this action chain remains incomplete as clinical decision support.