Mini-Mental State Examination Changes and Prediction of Cognitive Recovery after Stroke Following Cognitive-Motor Intervention.

Abstract

Objective: To evaluate the impact of cognitive-motor intervention on cognitive function in patients with stroke and explore predictive factors for clinically meaningful improvement.

Study design: An observational study. Place and Duration of the Study: Department of Rehabilitation Medicine, Anshun People's Hospital, Guizhou, China, from January 2023 to June 2025.

Methodology: Two hundred patients with cerebral infarction underwent a 3-week standardised programme combining structured cognitive training with motor exercises. MMSE scores were recorded at baseline and at 2, 4, and 8 weeks. Cognitive changes were analysed longitudinally using a linear mixed-effects model, and predictive models were tested for clinically meaningful improvement.

Results: MMSE scores improved significantly over time, with the mean MMSE increasing from 22.3 at baseline to 26.3 at week 8 (p <0.001). The main effect of group was not significant (p = 0.165); however, a significant time × group interaction (p <0.001) indicated that patients in the TMS group exhibited more rapid and greater improvement than controls. In predictive modelling, the random forest (RF) model achieved the best overall performance, with an AUC of 0.758 and an F1-score of 0.716, outperforming the other algorithms. XGBoost showed relatively weaker performance, whereas SVM demonstrated a slight advantage in specificity.

Conclusion: Cognitive-motor intervention provides sustained cognitive benefits after stroke. RF modelling may assist in the early identification of patients with a high potential for recovery, thereby supporting individualised rehabilitation.

Key words: Stroke, Cognitive-motor intervention, Transcranial magnetic stimulation, Mini-mental state examination, Machine learning, Rehabilitation, Neuroplasticity, Prediction.