Objective: To validate an AI-enhanced decision support tool to assist health care professionals (HCP) in distinguishing between patients with Parkinson’s disease (PD) and those with other tremor disorders.

Background: There are still many challenges in diagnosing PD, partly because other movement disorders present with similar symptoms, including tremor. It is particularly difficult to distinguish between PD and other tremor disorders at the early stages of PD, which is the cause of 48% misdiagnoses. Assessing the patients’ tremor and other subtle movement symptoms in an objective way in clinical practice will aid the HCP in the differential diagnosis of PD and other movement disorders with tremor.

Method: We assessed movement symptoms in 76 patients with movement disorders (49 PD, 27 non-PD). Data were collected from patients through drawing and writing tasks on a tablet with a proprietary multi-sensor digital pen (NeuroMotor Pen^TM) designed to measure and analyse neuromotor function. We extracted and analysed, partly by using Principal Component Analysis, 243 features quantifying multiple symptoms of movement abnormality. Those features and principal components were then used to train a classifier to distinguish between PD and non-PD.

Results: Our analysis revealed a set of features which can serve as biomarkers for differential diagnosis between PD and non-PD patients. Additionally, one of the principal components was found to be very informative in distinguishing between the two groups. A classifier was trained on a combination of features and tested using leave-one-out cross-validation, achieving 80.3% accuracy (70.4% specificity, 85.7% sensitivity).

Conclusion: The NeuroMotor Pen^TM shows significant potential in assisting the HCP in the differential diagnosis between PD and other tremor disorders. The accuracy level of the classifier is comparable to current diagnostic protocols, which typically require multiple visits over a period of months and more expensive neurophysiological tests and an improvement over the accuracy at the first visit. These encouraging results are currently under further validation in blind testing in a referred clinical cohort. Successful translation of the device into clinical pathways would enable cost and time to diagnosis benefits, as well as patient benefit.

« Back to 2023 International Congress

MDS Abstracts - https://www.mdsabstracts.org/abstract/a-digital-differential-diagnostic-aid-for-parkinsons-disease-and-other-tremor-disorders/