Flexible Wireless Smart Tremor Detection System

As global populations age, tremor-related symptoms have become increasingly prevalent in various neurological disorders. Traditional clinical assessments are limited by sporadic measurements and subjective evaluation, making it difficult to capture the full spectrum of a patient’s tremor throughout daily activities. To address these challenges, we have developed a g flexible, skin-conformable device equipped with tri-axial accelerometers and gyroscopes. The device wirelessly transmits motion data via Bluetooth to a smartphone application. Our processing pipeline extracts 264 time and frequency domain features from the raw sensor streams. After feature selection and cross-validation, we trained and compared three classifiers, Linear Discriminant Analysis (LDA), Logistic Regression, and AdaBoost, for tremor detection. In preliminary trials with six healthy volunteers simulating tremor through controlled muscle stimulation, the LDA classifier achieved the highest performance, with 68.3 % accuracy and an F1 score of 66.5 %. These results demonstrate the feasibility, comfort, and reliability of the flexible patch, as well as the promise of machine learning for automatic tremor detection. This work lays the foundation for future large-scale clinical validation and personalized remote monitoring solutions.