Cluster Analysis of Different Impaired Cognitive Domains in Patients With Post-Stroke Cognitive Impairment

A cluster analysis was conducted to classify the 7 different cognitive domains affected by PSCI patients, to explore the correlation and similarity between cognitive domains and provide a basis for targeted intervention.

We collected demographic and disease-related data from 724 PSCI patients. We used Python 3.8 software to perform K-means clustering and hierarchical clustering on the 7 cognitive domains assessed by the MoCA scale, and used the silhouette coefficient to determine the optimal number of clusters k.

The results of K-means clustering and hierarchical clustering show that the 7 dimensions of MoCA can be grouped into 2 categories. Cluster 1 scored lower in the cognitive areas of visual space and executive function, attention, language, abstraction, and delayed recall, whereas cluster 2 had higher scores in the naming and orientation domains. The scores in all cognitive domains of cluster 1 are lower than those of cluster 2, indicating severe cognitive impairment. Compared with cluster 2, the subjects in cluster 1 have poor physical health, living conditions, economic status, and social support ability.

The 7 dimensions of MoCA can be divided into 2 categories. In clinical practice, health care professionals should pay special attention to the severity of the patient's condition, the affected area, and individual differences, and develop precise and personalized treatment plans to improve the patient's cognitive function and quality of life.