github.com/terminalai/GaitMonitoringForParkinsonsDiseasePatients ↗

About terminalai/GaitMonitoringForParkinsonsDiseasePatients

This research project focuses on developing a wearable system to detect freezing of gait, a serious symptom affecting Parkinson's disease patients where the legs suddenly stop moving despite the patient's intention to walk. The team analyzed acceleration data from inertial measurement units placed on patients' thighs to identify which motion parameters could best predict these freezing events, ultimately aiming to improve safety and quality of life for affected individuals.

The project employs signal processing algorithms to extract a freeze index from three-axis acceleration data, then tests various support vector machine models to classify freezing versus normal gait. The researchers used the publicly available DAPHNet dataset containing recordings from ten Parkinson's patients with varying disease severity, performing their analysis through Google Colaboratory with GPU acceleration. Their findings indicated that the combination of vertical and horizontal lateral acceleration parameters provided the most accurate classification using a linear kernel SVM model.

The final deliverable is a lightweight prototype built on an Arduino Nano 33 BLE microcontroller with an integrated nine-axis motion sensor, designed to be worn on the thigh using an elastic strap. The device can detect freezing events in real time and trigger an LED indicator. While the current system demonstrates proof of concept, the team notes that future work should involve testing with actual Parkinson's patients, incorporating additional sensors beyond accelerometers for improved sensitivity, and integrating biofeedback mechanisms such as audio alerts or smartphone notifications to caregivers.