Related Experiment Video
Updated: Aug 6, 2026
08:07
Traditional Trail Making Test Modified into Brand-new Assessment Tools: Digital and Walking Trail Making Test
Published on: November 23, 2019
Machine learning based digital assessment of mild cognitive impairment using mouse trajectories during the trail
Gustavo E Juantorena1, Gianluca Capelo1,2, Betsabe D Leon Vallejos3
1Laboratorio de Inteligencia Artificial Aplicada, Instituto de Ciencias de la Computación, Facultad de Ciencias Exactas y Naturales, Universidad de Buenos Aires - CONICET, Buenos Aires, Argentina.
Scientific Reports
|July 21, 2026
Summary
Digital neuropsychology enhances assessments using computational methods. A computerized Trail Making Test (cTMT) using mouse movements shows promise in identifying mild cognitive impairment and predicting neuropsychological scores.
Area of Science:
- Neuroscience
- Computational Psychology
Background:
- Digital neuropsychology aims to enhance traditional assessments with computational methods.
- The Trail Making Test (TMT) is a key tool for evaluating executive functions.
- Mild cognitive impairment (MCI) diagnosis benefits from accurate neuropsychological assessments.
Purpose of the Study:
- To develop and validate a computerized Trail Making Test (cTMT) that captures high-resolution mouse trajectory data.
- To assess the utility of digital mouse features for discriminating between individuals with and without mild cognitive impairment.
- To determine if cTMT features can predict performance on standard neuropsychological tests.
Main Methods:
- Implemented a cTMT preserving the original TMT structure, recording detailed mouse trajectories.
- Developed the NeuroTask Python library to extract features from cursor time series data.
- Compared demographic, digital (mouse features), and combined models using nested cross-validation and permutation tests in 74 older adults (41 with MCI, 33 controls).
Main Results:
- Demographic models showed modest discrimination (AUC=0.56).
- Digital mouse features from the cTMT significantly improved classification performance (AUC=0.67).
- Combined demographic and digital features achieved an AUC of 0.70, approaching the diagnostic battery's performance (AUC=0.76).
- Regression analyses predicted four out of seven neuropsychological scores (MMSE, Digit Symbol, TMT-A, TMT-B).
Conclusions:
- Fine-grained mouse-movement features from the cTMT offer valuable information for classifying mild cognitive impairment.
- The cTMT has the potential to augment traditional neuropsychological assessments and improve diagnostic accuracy.
- Computational analysis of mouse trajectories represents a promising avenue for digital neuropsychology research.
