A Neural Network Model of Risk and Protective Factors for Poor Sleep Quality in Healthcare Providers: Role of Aggression and Self-regulation

Background: Although the literature discusses the benefits of good sleep on physiological, psychological, and physical health, poor sleep quality is common among healthcare providers. The present study aimed to predict poor sleep quality among healthcare providers based on multiple risk factors (aggression, gender, and age) and the protective factor of self-regulation using a neural network model.Methods: Using multistage cluster sampling, a group of ۴۰۰ healthcare workers (۷۰% female, with an age average of ۳۲ years) from Kermanshah city in western Iran were selected for the cross-sectional study. Data were collected using the Pittsburgh Sleep Quality Index (PSQI), the Buss-Perry Aggression Questionnaire (BPAQ), and the Self-Regulation Questionnaire (SRQ). A neural network model and receiver operating characteristic (ROC) curve were used for data analysis.Results: Four hidden units were found in a single hidden layer extracted using the current model. More than ۸۴% of the training and testing models accurately predicted good and poor sleepers. The neural network model's good predictive value was indicated by the Area under the ROC Curve (AUC=۰.۸۶۳). The results imply that self-regulation (۰.۳۰), anger (۰.۲۰), physical aggression (۰.۱۹), verbal aggression (۰.۱۱), hostility (۰.۱۰), age (۰.۰۶), and sex (۰.۰۵) have normalized importance values ranging from ۱۸% to ۱۰۰%, making them significant predictors of both the good and poor sleep subgroups.Conclusions: The present neural network-based algorithm, which considers the risk and protective factors of poor sleep quality, could be effectively used by healthcare providers.Background: Although the literature discusses the benefits of good sleep on physiological, psychological, and physical health, poor sleep quality is common among healthcare providers. The present study aimed to predict poor sleep quality among healthcare providers based on multiple risk factors (aggression, gender, and age) and the protective factor of self-regulation using a neural network model. Methods: Using multistage cluster sampling, a group of ۴۰۰ healthcare workers (۷۰% female, with an age average of ۳۲ years) from Kermanshah city in western Iran were selected for the cross-sectional study. Data were collected using the Pittsburgh Sleep Quality Index (PSQI), the Buss-Perry Aggression Questionnaire (BPAQ), and the Self-Regulation Questionnaire (SRQ). A neural network model and receiver operating characteristic (ROC) curve were used for data analysis. Results: Four hidden units were found in a single hidden layer extracted using the current model. More than ۸۴% of the training and testing models accurately predicted good and poor sleepers. The neural network model's good predictive value was indicated by the Area under the ROC Curve (AUC=۰.۸۶۳). The results imply that self-regulation (۰.۳۰), anger (۰.۲۰), physical aggression (۰.۱۹), verbal aggression (۰.۱۱), hostility (۰.۱۰), age (۰.۰۶), and sex (۰.۰۵) have normalized importance values ranging from ۱۸% to ۱۰۰%, making them significant predictors of both the good and poor sleep subgroups. Conclusions: The present neural network-based algorithm, which considers the risk and protective factors of poor sleep quality, could be effectively used by healthcare providers.

Keywords:

Aggression , Adolescents , Self-regulation , Sleep quality