Multiple-Wearable-Sensor-Based Gait Classification and Analysis in Patients with Neurological Disorders

Area of Science:

• Biomedical Engineering
• Neurology
• Rehabilitation Science

Background:

• Gait analysis is crucial for diagnosing and monitoring neurological disorders.
• Wearable sensors offer a promising non-invasive method for continuous gait assessment.
• Optimizing sensor placement and data analysis is key to accurate gait classification.

Purpose of the Study:

• To analyze the optimal placement of multiple wearable inertial measurement unit (IMU) sensors for classifying gait in neurological patients.
• To evaluate different feature sets and classification algorithms for distinguishing gait patterns.
• To determine the most effective sensor configuration for differentiating between healthy individuals, stroke patients, and other neurological disorder groups.

Main Methods:

• Seven IMU sensors were placed on participants at locations including the lower back (L5), thigh, shank, and foot.
• Temporal gait parameters were extracted and analyzed using various time-domain and gait temporal features.
• Classification was performed using Multilayer Perceptron (MLP) and Decision Tree (DT) algorithms with different sensor placements and feature combinations.

Main Results:

• A combination of time domain and gait temporal features with the MLP algorithm yielded superior classification accuracy.
• The sensor placed on the shank demonstrated higher accuracy compared to other placements (L5, foot, thigh).
• Shank sensor placement achieved 89.13% testing accuracy using the Decision Tree (DT) classifier.

Conclusions:

• Wearable IMU devices can effectively differentiate gait patterns across healthy individuals and patients with stroke or other neurological disorders.
• The combination of time domain and gait temporal features, along with shank sensor placement, provides the most favorable results for gait classification.
• This approach holds potential for improved diagnosis, monitoring, and rehabilitation strategies for neurological conditions affecting gait.