Sex-Specific Infarct Volume Associations and Early Prediction of Language Impairment Progression Following Stroke Surgery: A Network Approach

Guo-Dong Li¹Zhong-Hai Tao¹Ye-Kai Chen²Zhou-Kang Wu³Zi-Ning Zhu⁴Baihui Luo⁵Yu-Han Chen⁶Shan-Shan Ma⁶Bing Fu¹Hui Zheng^7*

• ¹The Second People's Hospital of Lianyungang, Lianyungang, China
• ²Zhejiang Normal University School of Psychology, Jinhua, China
• ³University of Science and Technology of China School of Humanities and Social Sciences, Hefei, China
• ⁴Zhenjiang Mental Health Center, Zhenjiang, China
• ⁵Guangzhou City Construction College, Guangzhou, China
• ⁶Kangda College of Nanjing Medical University, Lianyungang, China
• ⁷Shanghai Mental Health Center, Shanghai, China

Post-stroke language impairment affects nearly one-third of acute stroke patients or 30% of ischemic stroke patients, yet predicting its progression remains challenging due to multifactorial recovery processes. This study developed a predictive model for early language impairment deterioration following stroke surgery, integrating clinical, biochemical, and neuroimaging data. We enrolled 164 stroke surgical patients (China, 2022–2025), collecting data on clinical scales (National Institutes of Health Stroke Scale [NIHSS], Glasgow Coma Scale [GCS], modified Rankin Scale [mRS], Activities of Daily Living scale [ADL], Alberta Stroke Program Early CT Score [ASPECTS]), biochemical markers (C-reactive protein, glycated hemoglobin), and CT-derived infarct volume and affected brain regions. Least Absolute Shrinkage and Selection Operator (LASSO) regression identified eight predictors, including infarct volume, preoperative GCS, and ADL. Model performance was assessed via receiver operating characteristic and decision curve analyses, achieving an area under the curve of 0.80 (95% CI: 0.719–0.876) with 73.3% accuracy. Graphical Least Absolute Shrinkage and Selection Operator (GLASSO) and Bayesian network analyses revealed preoperative GCS as the primary direct predictor and mechanistic hub, with infarct volume and other factors exerting indirect effects. Males had larger infarct volumes (22.72 vs. 15.19 mL in females), but this did not directly correlate with worse language impairment outcomes. This multimodal model, enhanced by network analysis, accurately predicts language impairment progression and highlights preoperative consciousness as a key mediator, supporting precision stroke rehabilitation by capturing complex predictor interrelationships.