Decision making methodology based on generalized confidence and interpretability of artificial intelligence recommendation

Abstract

The article examines the transition in medical diagnostics from traditional clinician-dependent methodologies to evidence-based approaches using artificial intelligence (AI). The primary objective of the research is to develop a decision-making methodology based on the integration of human decisions and AI-based recommendations, as well as the interpretability of AI results for humans. The proposed methodology involves the formation of decisions based on human intelligence (HI) and AI, the assessment of the utility of recommendations, and generation of a joint decision based on cumulative probability. The practical application of the methodology was demonstrated through an experiment involving the classification of non-medical images. The research findings underscore the importance of transparency, interpretability, and trust in AI results for the successful utilization of AI in healthcare. Figs.: 1. Refs.: 16 titles.