Artificial neural network and decision tree models of post-stroke depression at 3 months after stroke in patients with BMI ≥ 24

Abstract

ObjectivePrevious studies have shown that excess weight (including obesity and overweight) can increase the risk of cardiovascular, cerebrovascular and other diseases, but there is no study on the incidence of post-stroke depression (PSD) and related factors in patients with excessive weight. The main purpose of this study was to find related factors of PSD at 3 months after stroke in patients with excessive weight and construct artificial neural network (ANN) and decision tree (DT) models. MethodsThis is a prospective multicenter cohort study (Registration number: ChiCTR-ROC-17013993). Five hundred and three stroke patients with Body Mass Index(BMI) ≥ 24 were included in this study. The diagnostic criteria of PSD is according to the Diagnostic and Statistical Manual of Mental Disorders, 5th edition (DSM-V) diagnostic criteria for depression due to other medical conditions and the HAMD-17 scores > 7 at 3 months after stroke was used as the primary endpoint. The χ2 test, Mann-Whitney U test or t-test were used to check for statistical significance. ResultsOur study found that sleeping time < 5 h, CHD, physical exercise, BI score, N dimension(EPQ) and subjective support(SSRS) were associated with PSD in patients with excessive weight. Physical exercise(odd ratio [OR] = 0.49, p = 0.001, 95%CI [confidence interval]: 0.32–0.75) and BI score(OR = 0.99, p < 0.001, 95%CI: 0.98–0.99) were protective factors; sleeping time < 5 h(OR = 2.86, p < 0.001, 95%CI: 1.62–5.04), CHD(OR = 2.18, p = 0.018, 95%CI: 1.14–4.15), N dimension(OR = 1.08, p = 0.001, 95%CI: 1.03–1.13) and subjective support(OR = 1.04, p = 0.022, 95%CI: 1.01–1.07) were risk factors. ConclusionThis study found several factors related to the occurrence of PSD at 3 months in patients with excessive weight. Meanwhile, ANN and DT models were constructed for clinicians to use.