Development of a Neurodegenerative Disease Gait Classification Algorithm Using Multiscale Sample Entropy and Machine...

Area of Science:

• Biomedical Engineering
• Neurology
• Data Science

Background:

• Neurodegenerative diseases (NDD) prevalence is increasing, necessitating advanced screening methods.
• Gait abnormalities are a known symptom of NDD, suggesting gait analysis as a viable screening approach.
• Current NDD screening methods can be invasive or costly.

Purpose of the Study:

• To develop and validate a machine learning-based algorithm for classifying NDD using gait force signals.
• To employ multiscale sample entropy (MSE) and statistical features for enhanced gait signal analysis.
• To assess the algorithm's performance across different NDD types and healthy controls.

Main Methods:

• Gait force (GF) signals were preprocessed, including differentiation and segmentation into various time windows.
• Feature extraction involved calculating statistical and multiscale sample entropy (MSE) values from GF signals.
• Data imbalance was addressed using the synthetic minority oversampling technique (SMOTE), with Support Vector Machine (SVM) and k-nearest neighbors (KNN) as classifiers.

Main Results:

• The algorithm achieved high classification accuracies, with KNN demonstrating superior performance under a 10-second time window.
• Specific accuracies included 99.90% for HC vs. PD, 99.80% for HC vs. HD, 100% for HC vs. ALS, and 99.68% for the four-class classification (HC vs. PD vs. HD vs. ALS).
• The developed method proved effective in distinguishing between various NDDs and healthy controls.

Conclusions:

• A robust NDD classification algorithm based on gait force, MSE, and machine learning has been successfully developed.
• The findings highlight the potential of gait analysis as a non-invasive and accurate tool for NDD screening and diagnosis.
• This approach offers a promising avenue for early detection and management of neurodegenerative diseases.