Developing a comprehensive malnutrition prediction model for the elderly in nursing homes.

Purpose: Malnutrition among elderly nursing home residents represents a critical public health challenge, particularly in rapidly aging societies such as China. This study aimed to develop and validate a predictive model for malnutrition risk tailored to this vulnerable population.

Methods: We analyzed clinical data from 1,023 elderly individuals (aged ≥ 65 years) across 26: nursing homes in Wuhan, China (March-October 2023). Participants were randomly divided into model-building (70%, n = 716) and internal validation cohorts (30%, n = 307). LASSO regression and logistic regression identified key predictors, and a nomogram was constructed. Model performance was assessed via AUC, calibration curves, and decision curve analysis (DCA).

Results: The malnutrition incidence was 46.37%. Five predictors were significant: feeding method (OR = 2.89, 95% CI: 1.75-4.76), dental status (OR = 0.56, 95% CI: 0.37-0.86), physical inactivity (OR = 1.75, 95% CI: 1.09-2.80), Barthel Index (OR = 0.96 per 10-point decrease), and anemia (OR = 1.91, 95% CI: 1.10-3.30). The model showed excellent discrimination (AUC = 0.90, 95% CI: 0.85-0.94) and calibration (mean absolute error = 0.026). DCA indicated clinical utility across threshold probabilities (2-97%).

Conclusion: This nomogram provides a robust tool for malnutrition risk stratification in nursing homes. Future studies should validate its generalizability across diverse populations and regions.