The Application of Machine Learning in Sports Medicine: A Scoping Review

Abstract

The application of artificial intelligence and machine learning (ML) in orthopaedic surgery is growing rapidly due to the exponential increase in health care data. In sports medicine, ML is increasingly being utilized to capture complex, nonlinear relationships within clinical data sets to aid in diagnosis and prognosis. This scoping review aimed to (1) identify the current applications of ML within sports medicine, (2) identify current strengths and limitations of ML with regard to clinical utility, and (3) identify future directions for ML and its potential applications in sports medicine. Scoping review; Level of evidence, 3. Utilizing the PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) guidelines, with the extension for scoping reviews, a comprehensive literature search was conducted in Embase, Medline, and Cochrane databases to identify studies focusing on ML applications in sports injuries requiring orthopaedic consultation. The search was performed on November 23, 2025. The inclusion criteria encompassed human studies assessing ML for diagnostic or prognostic purposes. The exclusion criteria included reviews, non-English studies, and studies on predicting or preventing primary injury. Descriptive statistics were used. A total of 55 studies were included out of 15,815 studies found on initial search. The most investigated surgeries were anterior cruciate ligament injuries/reconstruction and hip arthroscopy. Tree-based algorithms were the most frequently utilized models. ML was primarily applied to predict postsurgical resource utilization, clinical milestones, surgical survivorship, and biomechanical parameters. Limitations included a lack of external validation and clinical significance, the risk of model overfitting, and the “black box phenomenon.” This scoping review demonstrates that ML offers promising applications, but future research must prioritize external validation using large, high-quality data sets. Collaboration between clinicians and data scientists remains crucial to validate models that align with clinical needs, as does the input of physicians and patients alike in the shared decision-making process surrounding treatment.