JMIR Publications

ABSTRACT

Background:

The impact of surgical complications is substantial and multifaceted, affecting patients, families, surgeons and healthcare systems. Despite the significant advancement in artificial intelligence (AI), there still exists a notable gap regarding prospective implementation of AI models clinically using real time data to enhance the decision-making, and the ability to intervene proactively to reduce the risk of surgical complications.

Objective:

To this end, this scoping review aims to assess and analyze the adoption and use of AI models for preventing surgical complications. Furthermore, we aim to identify barriers and facilitators for the implementation and deployment at the bedside, where patient care occurs.

Methods:

A literature search was conducted using online databases including IEEE Xplore, Scopus, Web of Science, MEDLINE, ProQuest, PubMed, ABI, Cochrane electronic and registries for clinical trials from January 2013 to January 27, 2025. Study characteristics and algorithm development were extracted along with performance statistics (e.g. sensitivity, area under the operating curve). Preferred Reporting Items for Scoping Reviews (PRIMSA-ScR) was followed.

Results:

The search yielded a total of 275 of which 19 were included for data extraction. Out of these there were ten randomized controlled trials and nine prospective studies. In three studies a web version of a calculator for predicting postoperative complications was provided. None of the 19 studies reported that the model was in clinical use. Only one study reported the perceptions of clinicians in terms of using AI models.

Conclusions:

The literature demonstrates that AI models for surgical complications are not currently in use in routine practice. Barriers to use were identified as challenges of integration into daily clinical workflows and insufficient evidence regarding their clinical effectiveness.