Development and internal validation of a preliminary model for Helicobacter pylori infection: a cross-sectional study

Although Helicobacter pylori (Hp) infection is a major global public health challenge, clinically applicable tools for individualized risk stratification remain limited, particularly in geographically diverse and high-altitude populations.

To develop and internally validate a clinical prediction model for Hp infection using demographic, lifestyle, altitude-related, and clinical variables.

This multicenter cross-sectional study included 2034 participants recruited from three medical institutions in Qinghai Province, China, between November 2025 and March 2026. Hp infection status was determined using the ^13C-urea breath test. Participants were randomly assigned to a training set (n = 1423) and a validation set (n = 611). Candidate predictors associated with Hp infection in univariable logistic regression analysis (P < 0.05) were assessed for multicollinearity and subsequently entered into least absolute shrinkage and selection operator (LASSO) regression with 10-fold cross-validation. Predictors selected by LASSO were then included in a multivariable logistic regression model. A prediction model presented as a nomogram was subsequently developed. Model performance was assessed using receiver operating characteristic (ROC) analysis, calibration curves, and decision curve analysis (DCA).

Seven predictors were ultimately retained in the final model, including educational level, vegetable consumption, tooth brushing frequency, halitosis, family history of Hp infection, chronic gastritis, and high-altitude residence. The model demonstrated good discrimination, with area under the ROC curve (AUC) values of 0.774 in the training set and 0.756 in the validation set. Calibration analysis demonstrated good agreement between predicted and observed probabilities, with mean absolute errors of 0.018 and 0.037 in the training and validation sets, respectively. Decision curve analysis demonstrated consistent net clinical benefit in both sets.

We developed and internally validated a prediction model for Hp infection incorporating seven readily obtainable variables. The model demonstrated satisfactory discrimination, calibration, and clinical utility and may facilitate individualized risk stratification and targeted screening strategies, particularly in high-altitude and geographically diverse populations.