State dependence: Does a prior injury predict a future injury?. (May 2021)

• Record Type: Journal Article
• Authors: Stern, Benjamin D.; Hegedus, Eric J.; Lai, Ying-Cheng
• Is Part Of: Physical therapy in sport. Volume 49(2021)
• Page Start: 8
• Page End: 14
• Publication Date: 2021-05
• Subjects: Injury -- Prediction -- Sports -- Nonlinear; Sports physical therapy -- Periodicals; Sports injuries -- Patients -- Rehabilitation -- Periodicals; Athletic Injuries -- diagnosis -- Periodicals; Athletic Injuries -- therapy -- Periodicals; Physical Therapy -- Periodicals; Sports Medicine -- Periodicals; 615.82088796
• DOI: 10.1016/j.ptsp.2021.01.008
• Languages: English
• ISSNs: 1466-853X

Abstract

Abstract: The sports medicine literature is filled with associations between injury and causal factors. However, those results have been inconsistent. We're left wondering which of our athletes might need more attention and where our efforts might be best spent. Resistance to injury is the result of interaction between many variables. These variables are interdependent with dynamic relationships which can be sometimes correlated, at times anti-correlated and from time to time show no relationship with injury risk. Relationships we may have seen yesterday do not necessarily hold true for today and we should not use those to infer what will happen. This perspective piece builds on prior works and describes how the complex interaction between injury determinants presents in other systems, why determinants are not stable and instead vary over time due to internal and external forcing and why our prediction ability remains limited even when determinants are identified. Patterns built from frequent time series data in conjunction with nonlinear dynamical methods can offer us a new approach to thinking about injury prediction. Highlights: Relationships between variables in nonlinear systems are not constant and change with the state of the system. A factor can have different and sometimes opposite effects, helpful and harmful, on a system that is state dependent like the human athlete. Variables and relationships that play a role in injury risk may be overlooked by not considering state-dependence. It is possible to identify state dependence using nonlinear dynamical methods such as delay-coordinate embedding.