S-085 | A New Digital Neuropsychology Approach for Mild Cognitive Impairment: Computerized Trail Making Test with Integrated Hand and Eye Tracking

Disorders of the Nervous System
Author: Gustavo Juantorena | Email: gjuantorena@gmail.com

Gustavo Juantorena^1°, Waleska Berrios^2°, María Cecilia Fernández^2°,  Betsabe León^2°, Agustín Ibáñez^3°4°5°6°, Agustín Petroni^7°, Juan E. Kamienkowski^1°8°9°

^1° Laboratorio de Inteligencia Artificial Aplicada, Instituto de Ciencias de la Computación (FCEyN, UBA – CONICET)
^2° Departamento de Neurología Cognitiva, Hospital Italiano de Buenos Aires,
^3° Latin American Brain Health Institute (BrainLat), Santiago, Chile.
^4° Trinity College Dublin (TCD), Dublin, Ireland
^5° Global Brain Health Institute (GBHI)
^6° University of California San Francisco (UCSFA)
^7° Universidad de Gothenburg, Gothenburg, Suecia
^8° Departamento de Computación, Facultad de Ciencias Exactas y Naturales, Universidad de Buenos Aires, Argentina
^9° Maestría de Explotación de Datos y Descubrimiento del Conocimiento, Facultad de Ciencias Exactas y Naturales – Facultad de Ingeniería, Universidad de Buenos Aires, Argentina

In this study, we apply a novel computerized version of the Trail Making Test (c-TMT), designed to track hand and eye movements across multiple trials, addressing key limitations of the traditional pen-and-paper version. Our research involved adults diagnosed with mild cognitive impairment (MCI) and a matched healthy control group (HC). Using a standard computer mouse and eye-tracking hardware, we captured detailed task-related features and conducted various analyses to identify digital markers of cognitive function.
Participants also completed a comprehensive neuropsychological battery, including the Paper-and-Pencil TMT, Digit Symbol Test, Forward and Backward Digit Span, and the Clock Drawing Test. We explored the correlations between c-TMT measures and standardized executive function tests to uncover novel markers of executive function.
Our findings revealed statistically significant differences between the MCI and HC groups, not only in traditional behavioral measures (e.g., time to complete trials) but also in eye-movement-specific features such as scanpath length (i.e., number of fixations). Furthermore, we trained several machine learning algorithms on the hand and eye movement data to accurately classify MCI and healthy controls.
We advocate for complementary digital tools in neuropsychology to enhance the precision and effectiveness of cognitive assessments.