CNN-Based PD Hand Tremor Detection Using Inertial Sensors

Parkinson's disease (PD) affects tens of millions of people worldwide, simple and reliable engineering method assist doctors in diagnosis and treatment is necessary. In this letter, a convolutional neural network (CNN)-based PD hand tremor detection method is proposed. The wrist acceleration information of both PD patients with hand tremor and healthy subjects are acquired by wearable device with inertial sensor. Then a nine-layer CNN model is built to identify PD hand tremor. Through the comparing experiments and cross-validation, it is proved that this method can detect PD hand tremor symptoms effectively and has better performance than typical machine learning methods.

Automated summary

A convolutional neural network-based method for detecting Parkinson's disease hand tremors is proposed, which shows effective performance through experiments and validation.