Aggregate temporal injury co-occurrence across eight elite sports: An exploratory anatomical network analysis

Injury surveillance in elite sports emphasizes injury incidence and anatomical distribution, treating injuries as isolated events. However, musculoskeletal injuries frequently arise through interconnected biomechanical and kinetic-chain interactions across anatomical regions. Although network-based analytical approaches offer a novel framework for identifying latent interaction structures, application across elite sports remains limited. This retrospective observational study analyzed injury surveillance data collected over 10 months from boxing, athletics (running), bowling, badminton, para table tennis, short-track speed skating, fencing, and taekwondo. Injuries were classified into ten anatomical regions and aggregated at the sport-month level. Anatomical co-occurrence was defined as the joint presence of injuries in two regions within the same sport-month. Weighted undirected networks were constructed from aggregate temporal co-occurrences. Frequent anatomical pairings and community structure were examined using graph-theoretical analysis and Louvain community detection. The aggregate anatomical network demonstrated a densely connected structure. Frequent pairings included lumbar-shoulder, lumbar-wrist, shoulder-wrist, lumbar-knee, and knee-ankle. Community detection identified two communities within the pooled network. These patterns represent aggregate temporal associations and do not establish within-athlete anatomical or biomechanical relationships. Aggregate sport-month injury data exhibited recurrent temporal co-occurrence patterns across anatomical regions. These are exploratory population-level associations rather than evidence of within-athlete biomechanical coupling.