Abstract: With the integration of artificial intelligence into the medical field, intelligent auxiliary systems based on imaging devices and wearable technologies have been rapidly evolving, facilitating the early diagnosis of diseases. As a common cerebrovascular disorder in the elderly, cerebral small vessel disease (CSVD) has an insidious onset and slow progression, significantly impairing patients' quality of life. However, challenges such as limited data availability and class imbalance hinder the development of effective predictive models. Machine learning techniques are employed in this paper to investigate the intelligent prediction of CSVD. First, key risk factors are identified through feature selection. Then, the SMOTE-Tomek Links method is used to address the issue of data imbalance. Finally, a precise risk prediction model for CSVD is developed and validated using five-fold cross-validation. Experimental results demonstrate that the proposed method can significantly improve accuracy, recall, and area under curve (AUC) under conditions of imbalanced limited-sample data, providing an effective solution for the early prediction and diagnosis of CSVD.