Explainable AI for Medical Diagnosis: A Systematic Review and Meta-Analysis

The deployment of AI diagnostic systems in clinical settings requires not only high predictive accuracy but also interpretable explanations that clinicians can trust and act upon. This systematic review synthesizes 127 studies on explainable AI (XAI) for medical diagnosis published between 2018 and 2024, covering applications in radiology, pathology, genomics, and clinical decision support. We identify critical gaps: (1) evaluation of explanations rarely involves clinical validation, (2) most XAI methods are not designed for the temporal nature of longitudinal patient data, and (3) definition of "explanation quality" remains inconsistent across studies. We propose a unified XAI quality framework with five dimensions and highlight the most promising directions for clinically deployable interpretable AI.