Identification of Neurodegenerative Diseases Based on Vertical Ground Reaction Force Classification Using...

Area of Science:

• Biomedical Engineering
• Neurology
• Machine Learning

Background:

• Neurodegenerative diseases (NDDs) exhibit gait abnormalities affecting vertical ground reaction force (vGRF) signals.
• Early detection and monitoring of NDDs are crucial for effective patient management.
• Current diagnostic methods may lack the sensitivity to detect subtle gait changes.

Purpose of the Study:

• To develop and validate a novel deep learning algorithm for classifying neurodegenerative diseases (NDDs) using vGRF signals.
• To differentiate gait patterns between NDD patients and healthy controls (HC).
• To provide a tool for early NDD detection, treatment planning, and disease progression monitoring.

Main Methods:

• A deep learning approach was employed, involving signal preprocessing, feature transformation, and classification.
• vGRF signals were segmented into time windows (10, 30, 60 s).
• Continuous wavelet transform (CWT) converted time-domain vGRF to time-frequency spectrograms, followed by principal component analysis (PCA) for feature enhancement.
• A convolutional neural network (CNN) classifier was utilized and validated using leave-one-out cross-validation (LOOCV) and k-fold cross-validation (k=5).

Main Results:

• The proposed algorithm effectively differentiated gait patterns between HC subjects and NDD patients based on vGRF time-frequency spectrograms.
• The deep learning model demonstrated high accuracy in classifying NDDs.
• The analysis highlighted distinct vGRF signal variations in NDDs due to gait abnormalities.

Conclusions:

• A novel deep learning algorithm utilizing vGRF signals shows promise for neurodegenerative disease identification.
• The method offers a non-invasive approach for early detection and monitoring of NDDs.
• This technology can aid clinicians in managing NDDs more effectively.