Development and validation of a predictive model for severe insomnia in nurses: a machine learning–based approach (Preprint)

Abstract

BACKGROUND Insomnia is a prevalent condition among nurses, significantly impacting their health and job performance. Existing studies largely focus on correlating insomnia with various work-related and psychological factors without effectively predicting its severity. OBJECTIVE The aim of this study was to develop a predictive model for severe insomnia among nurses using machine learning techniques. METHODS Data were collected from February to October 2023 from four hospitals in China, using a multicenter cross-sectional study design. Insomnia severity was measured with the Insomnia Severity Index (ISI). Predictor variables were identified using the Boruta algorithm and Lasso regression. The predictive model was constructed using multivariable logistic regression and evaluated through five-fold cross-validation and external validation. Model performance was assessed in terms of area under the curve (AUC), Brier scores, and decision curve analysis. RESULTS A total of 1,969 valid questionnaires were collected, with 1,733 in the training set and 236 in the external test set. The prevalence of severe insomnia among nurses was 30%. The predictive model identified night shift frequency, challenge and hindrance–related stress levels, and career adaptability as significant predictors of severe insomnia. The model achieved an AUC of 0.723 (95% CI: 0.697–0.749) in the training set and 0.710 (95% CI: 0.638–0.783) in the external test set, indicating good discrimination ability. Brier scores of 0.181 and 0.184 were obtained on the training set and the external test set, respectively. The decision curve analysis showed greater net benefit at risk thresholds between 0.20 and 0.80. CONCLUSIONS This study developed a machine learning–based predictive model for severe insomnia among nurses. This model facilitates early identification and intervention, potentially improving nurses' sleep health and overall workplace well-being. Further research is recommended to explore additional predictive variables and refine the model for broader clinical application.