Machine learning analysis of 12,975 middle-aged and elderly Chinese participants identified nine key stroke predictors with sex-specific differences, and a nomogram model incorporating these factors exhibited better discrimination compared with other individual predictive factors.
Males had significantly higher stroke prevalence than females in the study population.
Eight machine learning algorithms identified nine key predictors of stroke risk in the study population.
High diastolic blood pressure, systolic blood pressure, total metabolic output, and Cystatin C were significantly associated with stroke risk in both males and females.
A nomogram prediction model developed from the identified risk factors demonstrated better discrimination than individual predictive factors alone.
Sex-stratified correlations among 27 health indicators revealed sex-specific patterns relevant to stroke risk in middle-aged and elderly Chinese individuals.
This research analyzed data from nearly 13,000 middle-aged and older Chinese adults followed over roughly nine years to identify which health measurements best predict stroke risk, and whether those predictors differ between men and women. Using eight different machine learning methods, researchers narrowed down 27 health indicators to nine key predictors, including blood pressure measures, blood fat and sugar indices, body size measurements, red blood cell characteristics, kidney function markers (Cystatin C), and physical activity output. Men were more likely to have strokes than women overall, and some risk factors were more strongly linked to stroke in one sex than the other. Four factors — high diastolic blood pressure, high systolic blood pressure, high total metabolic output, and high Cystatin C — were significantly associated with stroke risk in both men and women. The researchers then built a nomogram, which is a visual scoring tool that combines multiple risk factors into a single risk estimate. This combined tool performed better at distinguishing who would go on to have a stroke than any single measurement alone. This research suggests that stroke prediction in older adults is improved when sex-specific differences in risk factors are considered alongside a comprehensive combined risk score, rather than relying on any one measurement. The findings point to blood pressure control, metabolic health, and kidney function as broadly important targets, while also highlighting that men and women may need somewhat different monitoring approaches for optimal stroke prevention.
Huang X, Bao Q, Sun Y, Yang X, Zhang X. (2026). Analysis of sex-specific stroke risk factors in middle-aged and elderly Chinese population based on machine learning approach: A retrospective observational cohort study.. Medicine. https://doi.org/10.1097/MD.0000000000050391