Construction of a predictive model to predict the incidence of cerebral white matter lesions in young patients: a large retrospective study

Abstract

Abstract Objectives Recently, many studies have focused on the occurrence and development of cerebral white matter lesions in the elderly, while we focused on the risk factors leading to cerebral white matter lesions in younger patients and constructed a predictive model for predicting the incidence of lesions in younger patients. Materials and methods We studied patients younger than 60 years old who visited hospitals in Shin Takeo Hospital between april 1, 2016 and october 30, 2017. We randomised the data into test group and validation group. The model was developed by the test group and then validated by the validation group. The ROC curve, DCA curve (clinical decision curve) and clinical impact curve were used to verify the accuracy and clinical practical value of the model. Results 1050 patients were included in this study, including 363 patients with cerebral white matter lesions and 687 patients with non-cerebral white matter lesions. Multivariate analysis show that age(1.08(1.06, 1.11)), gender(1.69(1.27, 2.27)), carotid plaque score(1.19(1.07, 1.33)), diastolic blood pressure(1.02(1.01, 1.03)), Red_bp_med(1.50(1.01, 2.26)) were significantly associated with white matter lesions. The area under the curve (AUC) was 0.726 for the training group and 0.694 for the validation group. Our prediction model has high accuracy and clinical application value. Conclusions Age, gender, diastolic blood pressure, carotid plaque score and the use of antihypertensive drugs were independent risk factors for cerebral white matter lesions. The predictive model we constructed has good accuracy and clinical application value.