A study on predicting impaired fasting glucose risk in Chinese adults based on individual characteristics

This study aimed to develop a nomogram for early detection of impaired fasting glucose (IFG), predicting the 5-year risk in Chinese adults due to its link to various diseases.

This retrospective cohort study included 28,875 participants without IFG at baseline, randomly divided them to a training set and a validation set. We developed four predictive models-LASSO, full, stepwise, and MFP-ultimately selecting the LASSO model for nomogram development due to its simplicity and predictive performance. Four prediction model performance was assessed through ROC analysis, calibration curves, and decision curve analysis, with external validation using Shunde Hospital (n = 18,618) and NHANES (n = 2,038) dataset.

We developed a nomogram to predict the risk of IFG by incorporating parameters including age, body mass index (BMI), systolic blood pressure (SBP), fasting plasma glucose (FPG), and triglycerides (TG), which demonstrated performance with AUCs of 0.8167 and 0.8155 in the training and validation set, respectively. External validation achieved AUC 0.9665 (Shunde Hospital dataset) and 0.9171 (NHANES).

Our nomogram provides a personalized, validated approach for assessing 5-year IFG risk in Chinese adults, offering a practical screening tool for primary healthcare and resource-constrained environments.