Assistant Professor Lehigh University, United States
Introduction: Lower extremity injuries, notably those affecting the anterior cruciate ligament (ACL), stand as a pervasive challenge within the realm of medicine, particularly evident in high-impact sports like football and soccer. These injuries not only impede an immediate playing capability but also harbor long-term implications for overall performance and career longevity. ACL tears alone constitute a significant portion of sports-related injuries, approximately 68.3 per 100,000 person-years among individuals aged 10 to 64 years. Despite advances in both surgical technique and rehabilitation protocols, there still lacks objective measurement data for accurately predicting return-to-play time and remains an elusive task.
Wearable technology allows for the objective measurement of physiological biomarkers giving insight into a subject’s well-being. Devices such as Moxy Muscle Oxygen Monitor and Polar H10 heart rate monitor offer valuable insight into an athlete’s internal response to rehabilitation protocols. Complementing these objective measures, subjective assessments of factors such as rate of perceived exertion (RPE), soreness, and sleep quality provide additional layers in understanding experiences during recovery. With the recent advances in artificial intelligence (AI), it offers promise in managing various injuries. Analyzing objective physiological metrics from wearables and subjective assessments, machine learning can generate accurate predictive models for return-to-play timelines. These models identify patterns and enable personalized rehabilitation protocols, enhancing efficiency and outcomes in medical practices. Through the integration of wearables, AI, and subjective assessments the precision and efficacy of rehabilitation protocols can be optimized, benefiting patient populations.
Materials and
Methods: This IRB-approved study (#2113291-4) investigates the post-surgery rehabilitation progress of two Division 1 football athletes following anterior cruciate ligament (ACL) reconstruction surgery. Physiological parameters, including muscle oxygen saturation (SmO2) and heart rate, were tracked using the Moxy sensor and Polar H10 chest strap, respectively. Subjective metrics such as rate of perceived exertion (RPE), soreness, and sleep quality were recorded through standardized forms administered throughout the rehabilitation process.
Data analysis involved employing RStudio and Python for comprehensive statistical analysis, including Shapiro-Wilks test to check for normal distribution of data. Subsequently, Spearman's correlation plots were generated to explore potential relationships between subjective metrics and physiological parameters. Additionally, the Mann-Whitney U test was conducted to determine statistical differences in SmO2 between the surgical and contralateral legs.
In Python, the scikit-learn library was employed to develop a predictive machine learning model for athletes' return-to-play time post-ACL reconstruction surgery, focusing on decision trees and regression analysis. Leveraging the created dataset including features such as difference in SmO2, change in heart rate, and subjective metrics such as RPE, soreness and sleep quality, alongside general historical recovery data, a model was created to forecast the individual recovery timelines. Further validation and generalization are necessary for practical application in sports medicine settings.
Results, Conclusions, and Discussions: The statistical analysis results indicate that after 6 months, there is no significant difference in the muscle oxygen saturation (SmO2) levels between the surgical and contralateral leg (P = 0.698). The correlation matrix reveals a moderate negative correlation of -0.489 between the rate of perceived exertion (RPE) and hours of sleep, a weak negative correlation of -0.215 between soreness and RPE, and a negligible positive correlation of 0.017 between hours of sleep and soreness. Additionally, the data shows a consistent decrease in subjects' sleep duration over time, while soreness and RPE levels remained relatively stable.
The integration of wearable technology, subjective assessments, and machine learning presents a promising avenue for enhancing rehabilitation protocols and predicting return-to-play timelines for athletes recovering from ACL reconstruction surgery. This case study demonstrates the feasibility of tracking physiological parameters and subjective metrics throughout the rehabilitation process, providing valuable insights into the athletes' recovery progress. The lack of significant difference in muscle oxygen saturation levels between the surgical and contralateral legs after 6 months suggests a potential restoration of physiological balance post-surgery. Moreover, the observed correlations between subjective metrics such as rate of perceived exertion, soreness, and sleep quality underscore the intricate interplay between physiological and psychological factors during rehabilitation.
The findings of this study highlight the importance of personalized rehabilitation protocols tailored to individual characteristics and responses. By leveraging machine learning algorithms, predictive models can be developed to forecast return-to-play or return-to-daily living timelines, aiding practitioners in optimizing treatment strategies. Further validation and refinement of these models are imperative to ensure their efficacy and generalizability across diverse populations and injury severities. Future work will focus on feature selection to make the model more generalizable for populations post-surgery. Ongoing monitoring of progress using wearable technology and subjective assessments offers an objective measure of rehabilitation and can facilitate real-time adjustments to protocols, promoting faster and safer returns. With the integration of machine learning, there is potential to switch the paradigm in medicinal practices and improve outcomes for populations recovering from lower extremity injuries such as ACL tears.
Acknowledgements (Optional): The authors would like to offer their thanks to the Lehigh University Department of Athletics, Lehigh University Department of Sports Medicine, and Lehigh University Strength and Conditioning for their continued support.
