INTERPRETABLE MACHINE LEARNING IN HEALTHCARE: A SYSTEMATIC REVIEW AND RESPONSIBLE CLINICIAN–AI COLLABORATION FRAMEWORK

The integration of Artificial Intelligence (AI) in healthcare offers transformative potential for diagnosis, treatment, and decision-making.However, the widespread adoption of AI in clinical practice remains limited due to concerns about interpretability, transparency, and trust.This systematic review examines recent advancements in interpretable machine learning (IML) techniques within the healthcare domain, focusing on their role in enhancing clinician trust and improving collaborative decision-making.Literature from 2015 to 2025 was systematically reviewed across major scientific databases, following PRISMA guidelines.The review synthesizes stateof-the-art IML approaches-including model-agnostic explanation methods, inherently interpretable algorithms, and domain-specific visualization tools-and evaluates their clinical applicability, strengths, and limitations.Findings reveal that while IML can significantly enhance transparency, challenges such as inconsistent evaluation metrics, domain adaptation issues, and limited clinician engagement persist.