To gain the relationship between meteorological factors and the high-risk incidence of cardiovascular and cerebrovascular diseases (CVD) and to explore the preliminary application of machine learning methods in high-risk CVD prediction, based on CVD morbidity and mortality data from Anshun and Xingyi of Guizhou Province and the meteorological data during 2018—2019, we use machine learning algorithms to screen high-impact factors from the processed dataset, which includes bio-meteorological indices, deviation factors and variation factors. Four machine learning methods, i.e., Naive Bayes, K-Nearest Neighbors (KNN), Decision Tree, and Random Forest, are applied to investigate the relationship between meteorological factors and high mortality risk from CVD. The results show that: 1) High-risk CVD events are associated with extreme meteorological conditions, such as abnormal meteorological factors. In Anshun, over 50% of abnormal meteorological factors are extremely low, while in Xingyi, over 60% of meteorological factors are lower outliers, indicating significant heterogeneity in extreme factors. 2) The Decision Tree method has higher sensitivity in screening meteorological factors, with key factors primarily related to temperature and biometeorological indices, whereas the influence of pressure-related factors is weaker. 3) The predictive performance of machine learning varies regionally. In Anshun, mutual information feature selection combined with Random Forest produces optimal results, whereas in Xingyi, Decision Tree combined with KNN demonstrates the best performance overall. This study has identified a set of high-risk meteorological factors for CVD in mid-low latitude mountainous regions and validated the preliminary application of machine learning methods in regional high-risk CVD prediction. The findings could provide a scientific basis for the prevention and control of CVD in climate-sensitive areas.