Risk prediction of sleep disturbance in clinical nurses: a nomogram and artificial neural network model

Abstract

Background: Sleep disturbance occur among nurses at a high incidence. Aim: To develop a Nomogram and a Artificial Neural Network (ANN) model to predict sleep disturbance in clinical nurses. Methods: A total of 434 clinical nurses participated in the questionnaire, a cross-sectional study conducted from August 2021 to June 2022.They were randomly distributed in a 7:3 ratio between training and validation cohorts.Nomogram and ANN model were developed using predictors of sleep disturbance identified by univariate and multivariate analyses in the training cohort; The 1000 bootstrap resampling and receiver operating characteristic curve (ROC) were used to evaluate the predictive accuracy in the training and validation cohorts. Results: Sleep disturbance was found in 180 of 304 nurses(59.2%) in the training cohort and 80 of 130 nurses (61.5%) in the validation cohort.Age, chronic diseases, anxiety, depression, burnout, and fatigue were identified as risk factors for sleep disturbance. The calibration curves of the two models are well-fitted. The sensitivity and specificity (95% CI) of the models were calculated, resulting in sensitivity of 83.9%(77.5–88.8%)and 88.8% (79.2–94.4%) and specificity of83.1% (75.0–89.0%) and 74.0% (59.4–84.9%) for the training and validation cohorts, respectively. Conclusions: The sleep disturbance risk prediction models constructed in this study have good consistency and prediction efficiency, and can effectively predict the occurrence of sleep disturbance in clinical nurses.