JMIR Publications

ABSTRACT

Background:

Parkinson's Disease (PD) affects millions globally, causing motor function impairments. Early detection is vital, and diverse data sources aid diagnosis. We focus on lower arm movements during keyboard and trackpad/touchscreen interactions, which serve as reliable indicators of PD. Our study utilizes keystroke and mouse movement data, combined with patient PD status, to create a predictive model for detecting PD presence and severity.

Objective:

Through analyzing key timestamps, finger movements, finger movement speed, and hand movement patterns during keyboard input, we differentiate between individuals with PD and non-PD participants. This comparative analysis enables us to establish clear distinctions between the two groups and make predictions about the presence of the disease.

Methods:

Participants were recruited via email by the Hawaii Parkinson's Association (HPA) and directed to a web application for the tests. The application gathered participant demographics and conducted a series of tests to assess hand movement. Participants traced straight and curved ribbons using a trackpad or mouse to evaluate stability in two-dimensional hand movement. They also responded to on-screen prompts with keypresses to measure response times, accuracy, and unintended movements resulting in accidental presses.

Results:

Our study included 31 participants, 18 without PD and 13 with PD, and analyzed their lower limb movement data collected from keyboards and mice. From this dataset, we identified 28 relevant features and developed individual models based on these features. The models achieved F1 scores ranging from 0.55 to 0.8. Through extensive experimentation, we determined the optimal model configuration, which incorporated the top eight features. This model achieved a mean F1 score of 0.7722 (±0.1344) and a mean balanced accuracy of 0.65 (±0.1253) for predicting the presence of PD.

Conclusions:

This study confirms the viability of utilizing technology-based limb movement data to predict the presence of PD in individuals, demonstrating the practicality of implementing this approach in a cost-effective and accessible manner.