Artificial intelligence-enabled clinical decision support systems in preadmission testing: a scoping review of risk prediction, triage, and perioperative workflows (2020-2025).

Preadmission testing (PAT) is a critical step in perioperative care that supports risk stratification, triage, and optimization. Tools such as the American Society of Anesthesiologists Physical Status (ASA-PS) classification have limitations. This review mapped evidence on artificial intelligence–enabled clinical decision support systems (AI-enabled CDSS) and risk prediction tools in PAT and perioperative assessment, with particular attention to their implications for perioperative efficiency and patient safety. A scoping review was conducted following the Preferred Reporting Items for Systematic Reviews and Meta-Analyses extension for Scoping Reviews guidelines. PubMed, Embase, Scopus, and CINAHL were searched for English-language studies published between January 1, 2020, and August 1, 2025. Eligible studies applied artificial intelligence (AI) or machine learning (ML) to preoperative or PAT–related evaluation, risk prediction, triage, or decision support. Two reviewers independently screened all records. The review was preregistered on the Open Science Framework (DOI: https://doi.org/10.17605/OSF.IO/JKCRH ). The original registration described a broader “digital determinants of health” scope, which was refined to AI-enabled CDSS before data extraction. Fifty-six studies were included. Most were retrospective cohorts using imaging or electronic health record data. Radiomics and deep learning dominated oncologic prediction, while structured clinical and laboratory data informed models for anesthetic risk, transfusion, and postoperative complications. Natural language processing (NLP) predicted ASA-PS classification from preoperative text. Only a small number of prospective or randomized studies were identified. AI-enabled CDSS shows promise for perioperative risk prediction and PAT triage, but most applications remain at the proof-of-concept stage. When prospectively validated and embedded in perioperative workflows, these tools could streamline preoperative work-ups, reduce unnecessary testing and day-of-surgery cancellations, and support safer intra- and perioperative monitoring. Prospective, multicenter validation and real-world implementation studies are therefore needed before routine clinical use.