A novel classification method for balance differences in elite versus expert athletes based on composite multiscale...

Area of Science:

• Sports Science
• Biomechanics
• Human Movement Analysis

Background:

• Accurate balance assessment is vital for elite athletes but remains challenging.
• Traditional center of pressure (CoP) indicators lack consensus and may not capture complex postural control.
• Existing methods struggle to differentiate subtle balance variations in athletes.

Purpose of the Study:

• To investigate novel parameters for assessing balance control in elite athletes.
• To compare the effectiveness of traditional balance indicators with a new multidimensional approach.
• To evaluate machine learning algorithms for classifying athlete balance ability.

Main Methods:

• Collected center of pressure (CoP) data from elite athletes and freestyle skiers on varied surfaces and visual conditions.
• Calculated the Center of Mass Continuous Index (CMCI) and traditional time-domain features (AP/ML displacement, length, tilt).
• Applied Logistic Regression, SVM, Naive Bayes, and Ranking Forest classifiers, evaluated using ROC analysis.

Main Results:

• Traditional time-domain CoP features were insufficient for accurate balance discrimination.
• The Center of Mass Continuous Index (CMCI) demonstrated superior performance.
• The combination of CMCI and the Ranking Forest classifier achieved the highest sensitivity (0.95) and specificity (0.35).

Conclusions:

• A nonlinear, multidimensional approach using CMCI and machine learning is highly effective for assessing complex postural control in athletes.
• This method offers a more robust and accurate assessment of balance ability than traditional techniques.
• The findings suggest a promising direction for developing advanced balance assessment tools in sports science.