Multidisciplinary prediction of running-related injuries using machine learning
-
By
-
Han Wu
-
Katherine Brooke-Wavell
-
Michael R. Barnes
-
Zainab Awan
-
Sarabjit Mastana
-
Sam Allen
-
Richard C. Blagrove
-
February 6, 2026
Clinical Scorecard: Integrative Machine Learning Approaches for Predicting Running-Related Injuries
At a Glance
| Category | Detail |
|---|
| Condition | Endurance running-related injuries (RRI) |
| Key Mechanisms | Multifactorial risk factors including genetics, history, muscular strength, biomechanics, body composition, nutrition, and training |
| Target Population | Competitive endurance runners |
| Care Setting | Sports medicine and injury prevention monitoring |
Key Highlights
- Development of a machine learning-ready weekly RRI prediction dataset from 142 competitive endurance runners monitored over 12 months.
- Random forest models achieved the best predictive performance (AUC ~0.78), outperforming most other algorithms.
- Inclusion of a broader range of risk factors improved logistic regression model performance significantly.
Guideline-Based Recommendations
Diagnosis
- Utilize multidisciplinary risk factor assessment including genetic, biomechanical, and training data for individualized RRI risk evaluation.
Management
- Incorporate machine learning models, particularly random forest algorithms, to predict injury risk and guide preventive interventions.
Monitoring & Follow-up
- Prospective weekly monitoring of runners’ risk factors and injury status to enable timely prediction and management.
Risks
- Recognize the multifactorial nature of RRIs requiring comprehensive data collection to improve prediction accuracy.
Patient & Prescribing Data
Competitive endurance runners monitored weekly over 12 months
Machine learning models can stratify injury risk to inform personalized training adjustments and injury prevention strategies.
Clinical Best Practices
- Collect and integrate high-quality multidisciplinary risk factors for accurate injury risk prediction.
- Apply random forest machine learning models for superior predictive performance in RRI risk assessment.
- Use broader risk factor datasets to enhance logistic regression model accuracy where applicable.
- Maintain prospective and continuous monitoring to capture dynamic injury risk changes.
References
