Using Artificial Intelligence-Enhanced Sensing and Wearable Technology in Sports Medicine and Performance Optimisation

Swathikan Chidambaram, Yathukulan Maheswaran, Kian Patel, Viknesh Sounderajah, Daniel A. Hashimoto, Kenneth Patrick Seastedt, Alison H. McGregor, Sheraz R. Markar, Ara DarziSensors 2022 · OpenReview

TL;DR

This review explores the burgeoning intersection of Artificial Intelligence (AI) and wearable technology in sports medicine, aiming to optimize athlete performance and aid in injury management. By analyzing existing research, the paper highlights how AI can process complex sensor data to provide actionable insights across pre-event training, in-event strategy, and post-injury diagnosis and rehabilitation.

What Problem Does This Paper Address?

The paper addresses the growing need for sophisticated tools in sports medicine to enhance athlete performance and prevent injuries. While wearable technology has become ubiquitous, collecting vast amounts of physiological and kinematic data, the challenge lies in effectively processing and interpreting this data to derive meaningful insights. Traditional methods often struggle with the sheer volume and complexity of sensor data. This review aims to synthesize current research on how AI, particularly machine learning, can be leveraged with wearable sensors to overcome these limitations, offering a more personalized and data-driven approach to sports medicine and performance optimization.

Methods

This paper is a narrative review, meaning it synthesizes existing research rather than presenting new experimental results. The authors conducted a literature search across several major databases (MEDLINE, Embase, Scopus) using specific keywords related to AI, machine learning, wearable devices, and sports medicine. They included studies that used AI-based systems to process sensor data from wearable devices for diagnostic, treatment, or monitoring purposes in sports. The selected studies were then classified based on the phase of an athlete's journey: pre-event training, during events, and post-event. The quality of included studies was assessed using the QUADAS-2 tool.

Datasets

Not discussed in the paper. The paper reviews existing studies, and specific datasets used in those individual studies are not aggregated or presented as a primary focus of this review.

Key Insights and Ideas

  • AI as a "Digital Coach": AI, when integrated with wearable sensors, can act as a sophisticated "digital coach," providing real-time feedback and personalized insights that go beyond what traditional coaching or analysis can offer.
  • Holistic Athlete Monitoring: The combination of AI and wearables allows for a more holistic understanding of an athlete's condition, encompassing not just physical performance but also mental well-being (stress, anxiety, depression).
  • Predictive Power for Injury Prevention: AI algorithms can analyze complex patterns in physiological and kinematic data to predict the risk of injuries, enabling proactive interventions and reducing costly missed game time.
  • Personalized Rehabilitation: Post-injury, AI can help tailor rehabilitation programs by correlating sensor data with patient-reported outcomes, identifying individuals who may need more support to achieve meaningful recovery.
  • Data Challenges Remain: Despite the promise, significant challenges exist in data quality, standardization, transmission, and security, which need to be addressed for widespread adoption.

Key Results

The review highlights numerous studies demonstrating the efficacy of AI-enhanced wearables across various sports. For instance:

  • Injury Prediction: Machine learning models have shown the ability to predict injury risk with accuracies up to 94.6% for baseball and hockey players, outperforming conventional regression models.
  • Performance Optimization: AI has been used to analyze golf swings, weightlifting techniques, and basketball game data to identify areas for improvement.
  • Health Monitoring: Wearable devices with AI algorithms can detect significant physiological aberrations, such as those indicating potential cardiac events, with high accuracy (e.g., up to 83% for mental health characterization).
  • Rehabilitation Insights: Machine learning models have successfully correlated wearable sensor data with patient outcomes after procedures like total knee arthroplasty, aiding in the prediction of functional improvement.

Limitations and Future Work

The authors acknowledge several limitations and propose future directions:

  • Data Quality and Standardization: Obtaining high-quality, standardized data from wearables is challenging due to variations in spatial-temporal resolution and the need to integrate data from multiple devices. Future work should focus on developing more robust sensors and data processing techniques to handle noise, artifacts, and missing data.
  • Socio-economic Bias: The current reliance on wearable technology can create a socio-economic bias, as only those who can afford these devices can benefit. Future research should consider accessibility and affordability.
  • Data Security and Privacy: Ensuring the highest level of data security and privacy is a critical challenge.
  • Interoperability: Healthcare systems often lack interconnectedness between different databases. Future work needs to address device-to-device communication and integration into existing digital health platforms.
  • Patient Acceptability and User Adoption: A significant portion of wearable users stop using their devices. Future efforts must focus on educating the public about the benefits of AI in healthcare and addressing concerns about data exploitation and the potential loss of the human element in care.
  • Defining AI's Role: Clearly defining the specific roles, capabilities, and limitations of AI in wearable technology, with input from diverse stakeholders, is crucial for successful adoption.

Summary updated

BibTeX
@inproceedings{chidambaram2022using,
  author = {Swathikan Chidambaram and Yathukulan Maheswaran and Kian Patel and Viknesh Sounderajah and Daniel A. Hashimoto and Kenneth Patrick Seastedt and Alison H. McGregor and Sheraz R. Markar and Ara Darzi},
  title = {{Using Artificial Intelligence-Enhanced Sensing and Wearable Technology in Sports Medicine and Performance Optimisation}},
  booktitle = {Sensors},
  year = {2022},
  url = {https://openreview.net/forum?id=Wpaklr59QZ},
}

Loading related content...

Figure