Creation and verification of a predictive nomogram model for the incidence of social isolation among China's older population.

Objectives: To explore the risk factors associated with social isolation among the older adult in China, develop a nomogram model to forecast the risk, and evaluate its predictive accuracy.

Methods: An investigation was conducted into the demographic, socioeconomic, health, and health behavior aspects of the older adult population. Using logistic regression and backward stepwise analysis, a nomogram model was constructed to predict the risk of social isolation by screening independent risk factors.

Results: Social isolation was prevalent in 42.1% of Chinese older adults. Nomogram prediction models were created for the five screened variables, which included type of residence, health self-assessment, disability, depression, and sedentary hours. The nomogram model had an AUC of 0.734 (95%CI: 0.701-0.767) in the training cohort, and an AUC of 0.653 (95%CI: 0.580-0.725) in the validation cohort. The Hosmer-Lemeshow goodness-of-fit test revealed that there was a good fit ( > 0.05). DCA results showed that clinical intervention had a high net benefit in the older adult when the threshold probability was 20-85% for the training cohort and 30-65% for the control cohort.

Conclusion: Social isolation is a common issue for the older adult population in China. The prediction model using a nomogram for the older adult can efficiently detect and screen high-risk individuals for social isolation, forecasting its occurrence. The proposed nomogram may serve as a preliminary screening tool for social isolation risk but requires further optimization to improve accuracy. Future research should incorporate additional predictors or advanced modeling techniques to enhance clinical utility.