Smart Watch-Based Motion and Voice Analysis for Parkinson's Disease in Under-Resourced Settings

Parkinson's disease (PD) is the second most common neurological disorder in India. Accurate diagnosis of PD is limited due to the unavailability of sufficient resources. In this research, an IoT-based Parkinson's Smart Watch System that combines both the motion and voice analysis models to attain a real-time PD detection is being proposed. This system uses ESP32 microcontroller along with MEMS inertial sensors and microphones to capture and analyze the motor and voice patterns. The acquired voice data was evaluated using tuned K-Nearest Neighbor and Random Forest classifiers to achieve a prediction accuracy of 97.40% and 95%, respectively. The proposed watch also has the capability to detect falls based on the IMU sensor threshold values and gives SOS alerts powered by GSM modules and Twilio messaging systems to achieve immediate response during emergencies. This system also has additional features like individualized medication and meal reminders along with remote monitoring capacity to enhance patient care.