Cost-effectiveness of Artificial Intelligence applications in healthcare: A systematic review

Abstract

Objective With increasing emphasis on value-based healthcare and rising costs, it is essential to assess the economic impact of artificial intelligence (AI). This study systematically reviews the evidence on the cost-effectiveness of AI applications in healthcare. Methods A systematic search of PubMed, Scopus and Web of Science was conducted up to 21 August 2025. Studies that had full text and were peer-reviewed articles in English and reported formal economic evaluations, including cost-effectiveness, cost-utility or cost-benefit analysis of AI-based healthcare interventions, were included in this study. Key economic indicators such as incremental cost-effectiveness ratios (ICERs), quality-adjusted life years (QALYs) and disability-adjusted life years (DALYs) were extracted. Methodological quality was assessed using the Health Economics Consensus Criteria (CHEC-list). Data were qualitatively combined. Results A total of 26 studies met the inclusion criteria. Applications of AI in screening, diagnosis, treatment decision support and rehabilitation were assessed. Deep learning approaches were the most common approaches studied. Nine studies were classified as cost-effective, six as cost-neutral or somewhat cost-effective and eleven as not cost-effective. Overall methodological quality ranged from low to high, with considerable heterogeneity in model structure, perspective and time horizon. Conclusions AI in healthcare, particularly in screening and early diagnosis, shows promising cost-effectiveness. However, the strength of current evidence is moderate and highly context-dependent, and the results are sensitive to methodological assumptions, implementation costs, and healthcare system characteristics.