<![CDATA[Sports injury risk prediction using wearable sensor data and artificial intelligence has become increasingly relevant in athlete monitoring, yet the available evidence remains fragmented across sensor modalities, machine learning models, injury definitions, validation strategies, and implementation contexts. This study aimed to synthesize current evidence on artificial intelligence-driven wearable sensor data for sports injury risk prediction among athletes and to develop an integrative evaluation framework. A systematic literature review was conducted using Scopus as the main database, guided by the PRISMA flow and the PIOS framework. The initial search identified 136 records; all records were screened based on title, abstract, and metadata; 31 reports were assessed for eligibility; and 29 studies were included in the main extraction matrix. The data were analyzed using thematic synthesis with a narrative-integrative approach. The synthesis showed that injury risk was operationalized through biomechanical indicators, neuromuscular activation, physiological fatigue, movement quality, and training-load exposure. The dominant technologies included inertial measurement units, accelerometers, global positioning systems, electromyography sensors, wearable insoles, stretch sensors, heart-rate-related devices, and multimodal sensor systems. The applied models included logistic regression, random forest, support vector machine, XGBoost, long short-term memory, Transformer, temporal convolutional network–Transformer, spatial-temporal graph convolutional network, and multimodal fusion models. This review concludes that artificial intelligence and wearable sensor-based injury prediction should be evaluated as an integrated system connecting sensors, features, models, outcomes, validation, and implementation readiness.]]>
Copyrights © 2026
