Enhanced Binary Classification of Gait Disorders Using a Machine Learning Majority Voting Approach.

Area of Science:

• Biomedical signal processing
• Machine learning applications
• Gait analysis

Background:

• Gait disorders present a significant diagnostic challenge.
• Accurate identification of gait abnormalities is crucial for effective treatment.
• Existing methods for gait disorder classification have limitations in accuracy and scalability.

Purpose of the Study:

• To develop and validate a machine learning methodology for classifying healthy individuals and those with gait disorders.
• To compare the performance of various machine learning models using a merged dataset.
• To identify the most effective approach for accurate gait disorder identification.

Main Methods:

• Extracted key gait features from normalized ground reaction force data of 2435 subjects.
• Trained and optimized multiple machine learning models (SVM, Logistic Regression, Random Forest, Gradient Boosting, KNN, Bagging, Adaboost, Neural Network) using grid search.
• Evaluated model performance using accuracy, sensitivity, specificity, and F1 score via a repeated hold-out strategy (80% training, 20% testing).

Main Results:

• An ensemble model utilizing majority voting demonstrated superior performance compared to individual models.
• The majority voting ensemble model achieved an accuracy of 96.63%.
• This accuracy surpasses existing benchmarks in gait disorder classification.

Conclusions:

• Ensemble techniques, specifically majority voting, significantly enhance classification accuracy for gait disorders.
• The proposed machine learning methodology offers a scalable and cost-effective solution for gait disorder identification.
• This study contributes a robust tool for biomedical signal processing and clinical gait analysis.