On Thursday, August 7, 2025, Summer Research Intern Ford Love presented his poster “Neural Networks Extract Quantitative Features from MoCA Trails B Task to Detect Mild Cognitive Impairment in Parkinson’s Disease” to the UAMS DBMI faculty. Ford is a junior at the University of North Carolina at Chapel Hill and is studying Statistics and Music.
Parkinson’s Disease (PD) is a systemic neurodegenerative disorder causing both motor and cognitive impairment. The most recommended cognitive impairment screening exam is the Montreal Cognitive Assessment (MoCA) and is commonly used to determine the cognitive health of people with PD. Guided by PhD candidate Journey Eubank and Fred Prior Ph.D., Ford’s research focused on quantifying clinically meaningful features of patient performance from the MoCA Trail’s B task. In his presentation, Ford described his automated pipeline for preprocessing the MoCA exam by scaling and successfully classifying between three MoCA versions using ResNet50. Ford then isolated the Trails B task from the rest of the form using computer vision methods and extracted three clinically meaningful features—completeness of task, total number of errors, and mistakes in proper ordering of the task —achieving an F1 score of 0.916. Future work will focus on quantifying jitter in the drawn lines and any mistakes with corrections in the task. Ford’s research aims to quantitatively measure clinically meaningful features that may improve early detection of PD-MCI and support further refinements to cognitive assessments such as the MoCA.