Similarity Index Values in Fuzzy Logic and the Support Vector Machine Method Applied to the Identification of Changes...

Area of Science:

• Biomechanics
• Machine Learning
• Sports Science

Background:

• Proper supervision is crucial for correct resistance exercise execution.
• Identifying deviations in exercise form is essential for injury prevention and performance enhancement.
• Existing methods for evaluating exercise technique can be subjective and time-consuming.

Purpose of the Study:

• To propose and evaluate machine learning models for analyzing positional sequence patterns in barbell bicep curls.
• To compare the effectiveness of fuzzy logic (FL) and support vector machine (SVM) in identifying exercise execution changes.
• To assess the utility of Morisita-Horn similarity indices in conjunction with machine learning for movement pattern analysis.

Main Methods:

• Ten male volunteers performed barbell bicep curls with varying weights.
• Smartphone-based motion capture recorded joint positions and bar path in the sagittal plane.
• Fuzzy logic (FL) and Support Vector Machine (SVM) models were developed using similarity indices, deviation data, and expert evaluations.

Main Results:

• Significant deviations in vertical displacement were observed with a 50% body weight load (p < 0.002).
• Fuzzy logic (FL) models achieved superior performance (R² = 0.92, r = 0.96) compared to Support Vector Machine (SVM) models (R² = 0.81, r = 0.79).
• Expert evaluations showed over 70% agreement with statistical analysis of exercise execution.

Conclusions:

• Fuzzy logic (FL) modeling shows significant promise for automated assessment of exercise movement patterns.
• This approach can aid in detecting errors during resistance exercises and potentially enhance athletic motor performance.
• Machine learning, particularly FL, offers an objective and efficient tool for evaluating resistance exercise technique.