Predicting Aphasia Severity and Type Post-Stroke Using Structural MRI and Machine Learning

Summarise with VIRAT
Authors: Olga V. Buivolova
Cited by: 0

Summary

No summary yet. VIRAT can read this paper and write a plain-language summary of its aim, methods, key findings and limitations.

Summarise with VIRAT Free to try with your Science Catalogs account.

Abstract

Overview This study introduces a machine learning framework designed to predict post-stroke speech and language disorder profiles from structural MRI (sMRI) data. By operating within Luria's neuropsychologically grounded taxonomy, the work addresses critical gaps in the field, offering a theoretically informed approach to prognostication. Objectives The study pursued two primary objectives: 1) Prognostic Modeling: Develop and validate a robust machine learning model capable of predicting overall aphasia severity and syndrome types according to Luria's classification. 2) Neural Correlates: Identify the specific gray and white matter integrity metrics most critical to these predictions. Dataset The study utilized a large and unique cohort comprising 304 Russian-speaking individuals with chronic aphasia and dysarthria. The validation cohort comprised 41 participants. Quantitative metrics of gray and white matter integrity were extracted from structural MRI data and integrated with key demographic variables. The clinical-demographic data and GM and WM metrics extracted from sMRI are presented here.

Fields of Study & Subjects

Neurobiology of Language and Bilingualism Dementia and Cognitive Impairment Research Traumatic Brain Injury Research

Publication details

Journal / source
OSF Preprints (OSF Preprints)
Publication date
20 March 2026
Language
EN
Document type
Preprint