A Multimodal Predictive Model for High Imaging Burden in Cerebral Small Vessel Disease

Background Cerebral small vessel disease (CSVD) is a leading cause of vascular cognitive impairment. Early identification of patients with high CSVD imaging burden (composite score 3–4) remains challenging, as existing tools lack integration of neuroimaging with peripheral biomarkers. Methods This prospective study included 103 patients with CSVD, stratified into a low imaging burden group (score 0–2, n = 56) and a high imaging burden group (score 3–4, n = 47). Predictors were screened from multidimensional data (neuroimaging, blood markers, neuropsychological assessment) using LASSO regression. A multifactorial logistic regression model was constructed and visualized as a nomogram. The model underwent triple validation: calibration (Hosmer-Lemeshow test, Brier score), discrimination (ROC curve, AUC), and clinical utility (decision curve analysis). Results Four independent predictors of high CSVD imaging burden were identified: impaired language ability (OR = 3.41, 95% CI: 1.54–7.55; p = 0.002); high modified Fazekas score (OR = 0.06, 95% CI: 0.02–0.20; p < 0.001); lymphocytopenia (OR = 0.33, 95% CI: 0.13–0.87; p = 0.025); and elevated fibrinogen (OR = 0.32, 95% CI: 0.11–0.96; p = 0.042). The model showed excellent performance: AUC = 0.93 (95% CI: 0.88–0.98), high calibration accuracy (Brier score = 0.11), and significant net benefit across the 20%-80% threshold probability range. Conclusions This multimodal model integrates CSVD-specific neuroimaging (Fazekas score), peripheral immune-coagulation markers (lymphocyte count, fibrinogen), and a core cognitive domain assessment (language ability) to provide early risk stratification for high CSVD imaging burden, a key state associated with increased risk of cognitive decline.