Prospective Forecasting of Quadriceps Performance Post-ACLR in Collegiate Athletes

Area of Science:

• Orthopedics and Sports Medicine
• Biostatistics
• Rehabilitation Science

Background:

• Accurate prognosis of quadriceps function post-anterior cruciate ligament reconstruction (ACLR) is crucial for clinical expectations and interventions.
• Advanced probabilistic modeling, like Bayesian frameworks, can enhance recovery forecasting for personalized clinical decisions.

Purpose of the Study:

• To characterize quadriceps function recovery in collegiate athletes up to 14 months post-ACLR.
• To evaluate how the number of prior assessments influences the accuracy of Bayesian model predictions for recovery.

Main Methods:

• A cohort of 66 Division I collegiate athletes underwent serial isometric assessments (peak torque, rate of torque development, torque steadiness) from 1-14 months post-ACLR.
• Bayesian hierarchical B-spline models incorporating graft type and time post-ACLR were used, with cross-validation to assess predictive accuracy based on increasing observations.
• Root mean square error (RMSE) was used to quantify prediction accuracy.

Main Results:

• The most accurate model allowed recovery curves to vary by graft type and included athlete-specific random intercepts, achieving low RMSEs for predicting peak torque (0.109), rate of torque development (0.173), and torque steadiness (0.536) after just two assessments.
• On average, athletes required 10 months for peak torque (90%), 12 months for rate of torque development (85%), and 6 months for torque steadiness (0.5% of PT) to reach clinical targets with 95% certainty.

Conclusions:

• Bayesian hierarchical modeling provides a robust method for forecasting quadriceps recovery post-ACLR.
• Rate of torque development was identified as the metric taking the longest to achieve clinical targets.
• This predictive modeling approach serves as a proof-of-concept for individualizing patient recovery trajectories.