A nomogram incorporating nutritional risk for predicting 31-day unplanned readmission in adults with type 2 diabetes mellitus using LASSO regression

This is the authors' abstract. We don't add key points for this paper.

Objective To develop and validate a nomogram incorporating nutritional risk for predicting 31-day unplanned readmission in patients with type 2 diabetes mellitus (T2DM). Methods This retrospective study included 1,627 T2DM patients. A nomogram was constructed using least absolute shrinkage and selection operator (LASSO) regression on a training set ( n = 1,139) and an internal hold-out set ( n = 488). Predictors were selected from demographics, lifestyle, clinical, and nutritional parameters. Model performance was evaluated using ROC curves, calibration plots, and decision curve analysis. Results The 31-day unplanned readmission rate was 18.50%. Six predictors were identified: smoking history, peripheral vascular disease in patients with type 2 diabetes mellitus (PVD in T2DM), diabetes duration, glycated hemoglobin (HbA1c), nutritional risk and neutrophil-to-lymphocyte ratio (NLR). The nomogram showed good discrimination in both training (AUC = 0.846; 95% CI: 0.814–0.878) and internal hold-out set (AUC = 0.817; 95% CI: 0.761–0.873), with acceptable calibration. Conclusion In this single-center retrospective cohort, the nomogram demonstrated potential for predicting 31-day unplanned readmission in T2DM patients by integrating clinical, metabolic, inflammatory, and nutritional dimensions. Routine nutritional assessment at admission may help identify high-risk individuals; however, external validation in geographically and temporally diverse populations is necessary before clinical application.