Integrating Wearable Sensor Technology and Machine Learning for Objective m-CTSIB Balance Score Estimation.

Area of Science:

• Biomechanics
• Wearable Technology
• Machine Learning

Background:

• Balance assessment is crucial for adults at risk of falls and cognitive decline.
• Current methods can be costly and inaccessible.
• Objective and affordable balance assessment tools are needed.

Purpose of the Study:

• To develop an objective, affordable, and accessible method for balance assessment using wearable technology and machine learning.
• To correlate sensor data with Modified Clinical Test of Sensory Interaction on Balance (m-CTSIB) scores.
• To evaluate the efficacy of machine learning models in predicting balance performance.

Main Methods:

• Utilized a single inertial measurement unit (IMU) sensor to collect lumbar accelerometer and gyroscope data.
• Collected ground truth data from m-CTSIB tests performed on a force plate.
• Employed XGBOOST machine learning algorithm for data analysis.
• Implemented a subject-wise leave-one-out cross-validation technique.

Main Results:

• Achieved a 0.94 correlation between accelerometer data and m-CTSIB scores.
• Achieved a 0.90 correlation between gyroscope data and m-CTSIB scores.
• Demonstrated strong predictive accuracy for balance assessment.

Conclusions:

• Wearable technology combined with machine learning provides an objective and affordable method for balance assessment.
• This approach shows significant potential for remote monitoring and improved diagnosis of balance disorders.
• The findings can enhance the management of balance issues, improving quality of life and independence in adults.