Background: The mechanism(s) of cognitive impairment remains complex, making it difficult to confirm the factors influencing poststroke cognitive impairment (PSCI).
Objective: This study quantitatively investigated the degree of influence and interactions of clinical indicators of PSCI.
Methods: Information from 270 patients with PSCI and their Wechsler Adult Intelligence Scale (WAIS-RC) scores, totaling 18 indicators, were retrospectively collected. Correlations between the indicators and WAIS scores were calculated. Multiple linear regression model(MLR), genetic algorithm modified Back-Propagation neural network(GA-BP), logistic regression model (LR), XGBoost model (XGB), and structural equation model were used to analyze the degree of influence of factors on the WAIS and their mediating effects.
Results: Seven indicators were significantly correlated with the WAIS scores: education, lesion side, aphasia, frontal lobe, temporal lobe, diffuse lesions, and disease course. The MLR showed significant effect of education, lesion side, aphasia, diffuse lesions, and frontal lobe on the WAIS. The GA-BP included five factors: education, aphasia, frontal lobe, temporal lobe, and diffuse lesions. LR predicted that the lesion side contributed more to mild cognitive impairment, while education, lesion side, aphasia, and course of the disease contributed more to severe cognitive impairment. XGB showed that education, side of the lesion, aphasia, and diffuse lesions contributed the most to PSCI. Aphasia plays a significant mediating role in patients with severe PSCI.
Conclusions: Education, lesion side, aphasia, frontal lobe, and diffuse lesions significantly affected PSCI. Aphasia is a mediating variable between clinical information and the WAIS in patients with severe PSCI.
Keywords: Alzheimer’s disease; WAIS-R; cognitive impairment; predictor; stroke.