Patient centric trustworthy AI in medical analysis and disease prediction: A Comprehensive survey and taxonomy

Abstract

Artificial Intelligence (AI) integration in healthcare is revolutionizing medical analysis and disease prediction, enhancing diagnostic accuracy and patient care. However, with the growing adoption of AI, concerns surrounding trustworthiness, ethics, and transparency persist. This survey paper explores Trustworthy AI in healthcare, with a distinct focus on a patient-centric approach. By analyzing 132 relevant papers, we present a novel taxonomy across ten dimensions, emphasizing the criticality of safety, robustness, and patient trust. We highlight factors influencing trustworthiness and investigate the ethical frameworks guiding responsible AI deployment. A key contribution is the introduction of the Trustworthy AI Scoring System (TAI-SS), a novel framework to assess AI trustworthiness in healthcare, emphasizing ethics, privacy, and reliability. Case studies, such as AI-powered cancer diagnosis, demonstrate TAI-SS’s practical application. Additionally, we discuss transparency through Explainable AI (XAI) techniques and segmentation approaches. Our analysis underscores the importance of healthcare datasets and AI algorithms while recommending seven Trustworthy AI requirements and four ethical principles. This paper serves as a roadmap for AI-driven, patient-centric healthcare, offering insights for researchers, healthcare professionals, and policymakers.