Estimating Postural Stability Using Improved Permutation Entropy via TUG Accelerometer Data for Community-Dwelling...

Area of Science:

• Gerontology
• Biomedical Engineering
• Data Science

Background:

• Effective fall prevention programs require accurate identification of elderly individuals at risk.
• Current assessment methods may benefit from supplementary objective data.
• Early identification of fall risk is crucial for timely intervention.

Purpose of the Study:

• To develop a sensor-based methodology to supplement traditional fall risk assessments.
• To utilize feature selection, entropy analysis, and statistical analysis for improved fall risk classification.
• To compare sensor-derived fall risk predictions with clinical judgment and existing scales.

Main Methods:

• Application of logistic regression analysis to inertial sensor data.
• Employing permutation entropy and statistical feature analysis for risk assessment.
• Calculating the area under the receiver-operating characteristic curve (AUC) for performance evaluation.
• Comparing sensor-based classification with the short-form Berg balance scale.

Main Results:

• Permutation entropy and statistical features achieved AUC values exceeding 0.9, minimizing false positives.
• The weighted-permutation entropy/statistical features approach showed good agreement with the short-form Berg balance scale.
• Logistic regression effectively identified high-risk patients using inertial data and selected features.
• The proposed method offers a more accurate classification of fall risk in the elderly.

Conclusions:

• Sensor technology combined with entropy and statistical analysis provides a robust tool for fall risk assessment.
• This approach enhances the accuracy of identifying elderly individuals at high risk of falling.
• The methodology supports clinical decision-making for targeted fall prevention interventions.