From algorithm to clinic: confronting the barriers to artificial intelligence integration in personalized medicine

Purpose This study aims to critically assess the maturity and readiness of artificial intelligence (AI) applications in personalized healthcare and to identify the barriers hindering their clinical integration. Design/methodology/approach Adhering to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses guidelines, we conducted a comprehensive analysis of 1,290 peer-reviewed studies sourced from PubMed, Scopus, Google Scholar and IEEE Xplore, spanning the years 2019–2024. The ROBINS-I tool was utilized to evaluate bias, and following a rigorous full-text review, 15 studies met the inclusion criteria. A novel maturity model was developed to categorize the studies. Findings The vast majority of studies (13/15) are concentrated at Stage 1 (technical validation) and Stage 2 (clinical efficacy), with a complete absence of studies at Stages 4–5. None have achieved full integration into clinical workflows. We identify a critical “implementation chasm” caused by a trust deficit (black box problem), an equity crisis (algorithmic bias) and a governance gap (data privacy and regulatory challenges). Practical implications The field must shift its focus from technical performance to addressing the systemic barriers identified. We propose a maturity model to guide future research and development towards sustainable and equitable integration. Originality/value This review moves beyond cataloging benefits and challenges by introducing a maturity model and critically examining the implementation chasm. It provides a novel framework for advancing AI in personalized healthcare.