Miao Da,1 Shaoqi Mou,2 Guangwei Hou,3 Zhongxia Shen1 

1Department of Sleep Medicine Center, Huzhou Third Municipal Hospital, the Affiliated Hospital of Huzhou University, Huzhou, Zhejiang, People’s Republic of China; 2Department of Medicine, Wenzhou Medical University, Wenzhou, Zhejiang, People’s Republic of China; 3Department of Psychiatry, Yuyao Third People’s Hospital, Ningbo City, Zhejiang Province, People’s Republic of China

Correspondence: Zhongxia Shen, Department of Sleep medicine center, Huzhou Third Municipal Hospital, the Affiliated Hospital of Huzhou University, 2088 East Tiaoxi Road, Huzhou, Zhejiang, People’s Republic of China, Tel +860572-2132661, Email snowszx@sina.com

Objective: This study aimed to analyze the changes in insomnia characteristics among the general population and explore associated factors during the COVID-19 pandemic and post-pandemic periods.
Methods: A cross-sectional study was conducted using an anonymous online survey. Questionnaires were administered at two-time points (T1: March 1– 31, 2022; T2: March 1– 31, 2023), which included an Insomnia Severity Index (ISI) and questions related to sleep risk factors, including the COVID-19 pandemic, familial influences, work and study conditions, social activities, physical health, use of electronic devices before sleep, sleep environment, food intake and exercise before sleep, etc. Insomnia characteristics were compared at two points, with logistic regression testing associations with sociodemographic covariates and risk factors. Six machine learning models were employed to develop a predictive model for insomnia, namely logistic regression, random forest, neural network, support vector machine, CatBoost, and gradient boosting decision tree.
Results: The study obtained 2769 and 1161 valid responses in T1 and T2, respectively. The prevalence of insomnia increased from 23.4% in T1 to 34.83% in T2. Univariate analyses indicated the factors of the COVID-19 pandemic, familial influences, social activity, physical health, food intake, and exercise before sleep significantly differed in T1 (p< 0.05) between insomnia and non-insomnia groups. In T2, significant differences (p< 0.05) were observed between the two groups, including the factors of the COVID-19 pandemic, family structure, work and study conditions, social activity, and physical health status. The random forest model had the highest prediction accuracy (90.92% correct and 86.59% correct in T1 and T2, respectively), while the pandemic was the most critical variable at both time points.
Conclusion: The prevalence and severity of insomnia have worsened in the post-pandemic period, highlighting an urgent need for effective interventions. Notably, the COVID-19 pandemic and physical health status were identified as significant risk factors for insomnia.

Keywords: COVID-19, public, insomnia, risk factors, machine learning