Biomarkers.

Area of Science:

• Neurology
• Speech-Language Pathology
• Computational Linguistics

Background:

• Neurologic diagnostics require equitable access, prompting the development of the Mobile Assessment of Speech To Evaluate, Recognize, and Monitor Indicators of Neurologic health and Disease (MASTERMIND).
• The MASTERMIND Assessment of Naming subtest uses 10 pictures from a list of 45 words to assess naming abilities.

Purpose of the Study:

• To evaluate the feasibility of a remotely self-administered naming test for individuals with cognitive, linguistic, or communication impairments.
• To assess the sensitivity and specificity of performance measures (accuracy, error type, response time) for detecting aphasia and cognitive-communication difficulties.

Main Methods:

• The naming task was remotely administered to 138 controls and various patient groups (cognitively impaired, motor speech disorders, cognitive concerns, aphasia).
• Natural language processing (NLP) and advanced speech transcription (Whisper Large-v3) were used to analyze participant responses.
• Performance metrics were analyzed using statistical tests, with AUROC analysis evaluating diagnostic performance.

Main Results:

• Significant differences in accuracy, response time, and error types were observed across participant groups.
• While accuracy alone had limited predictive power for all patients versus controls, it showed better performance in distinguishing aphasia patients from other groups.
• Aphasia and motor speech disorder groups exhibited longer response times compared to controls and cognitively impaired participants.

Conclusions:

• The MASTERMIND Assessment of Naming is a scalable, feasible, self-administered tool leveraging NLP for neurologic health assessment.
• Response time, accuracy, and error analysis offer valuable insights for patient triaging and screening.
• Future research will focus on refining the tool's sensitivity, expanding sample diversity, and enhancing usability.