Laser-induced Graphene Sensor Interfaced with Machine Learning Classifiers for Hand Tremor Identification

Parkinson's disease is often diagnosed based on clinical signs, such as the description of a range of movement symptoms, and medical observations. Traditional diagnostic methods could be prone to subjectivity issues because they depend on the evaluation of subtle motions that might be difficult to define with the human eye. Hence, simple and reliable engineering methods are required for the early diagnosis of Parkinson's disease and for providing timely treatment to the tens of millions of patients who are affected by this disease. In this work, we developed a flexible sensor using laser-induced graphene to differentiate between natural finger bending and hand tremor. Measurements were recorded when the sensor was bent at various angles (30 degrees, 60 degrees, 90 degrees, 120 degrees, and 150 degrees) to mimic the natural finger-bending movements. Moreover, hand tremor was simulated by shaking the sensor vigorously. Our findings demonstrated a nearly linear relationship between the peak voltage measured across the laser-induced graphene sensor and the amount of bending. The degree of natural bending and hand tremors were multi-classified using a neural network machine learning classifier, showing an accuracy of 78.5% and an area under the curve of 0.97. The results of this study are promising in making an informed decision about differentiating Parkinson's-induced hand tremors from natural body movements and reducing misdiagnosis of Parkinson's.