Classifying Unstable and Stable Walking Patterns Using Electroencephalography Signals and Machine Learning...

Area of Science:

• Neuroscience
• Biomedical Engineering
• Machine Learning

Background:

• Analyzing gait stability using electroencephalography (EEG) is crucial for developing fall prevention systems.
• Real-time brain-computer interface (BCI) systems require accurate detection of gait instability from neural signals.
• Falls pose significant risks, necessitating advanced methods for early detection and intervention.

Purpose of the Study:

• To investigate the feasibility of using EEG signals for classifying stable and unstable gait patterns.
• To evaluate the effectiveness of various machine learning algorithms in detecting walking instability from EEG data.
• To develop EEG-based digital biomarkers for gait analysis and fall risk assessment.

Main Methods:

• Acquired 64-channel EEG signals from 13 healthy adults during four walking conditions (normal, medial-lateral perturbation, dual-tasking, visual feedback).
• Extracted digital biomarkers, including wavelet energy and entropies, from the EEG signals.
• Applied machine learning algorithms such as ChronoNet, SVM, Random Forest, gradient boosting, and LSTMs for classification.

Main Results:

• Machine learning models achieved classification accuracies ranging from 67% to 82% for different gait patterns.
• EEG-based digital biomarkers effectively differentiated between stable and unstable walking conditions.
• The study demonstrated the potential of EEG signals in identifying neural correlates of gait instability.

Conclusions:

• Classifying gait patterns using EEG signals and machine learning is feasible and accurate.
• EEG-based digital biomarkers show promise for real-time detection of unsteady gait.
• This research contributes to the development of BCI systems for fall prevention and injury reduction.