The clinical value of analyzing predictive models for gastric function and gastric organic lesions

To explore the clinical value of the predictive models for gastric function and gastric organic lesions.

Clinical data from 1,677 patients with digestive tract-related symptoms at the First Affiliated Hospital of Guangdong Pharmaceutical University between June 2015 and March 2023 were selected, including questionnaire responses, serum gastric function tests, gastroscopy, and gastroscopic pathology data. The above data were input into a gastric function prediction model based on question-and-answer [GFPSBQA] V1.0 and a gastric organic lesion prediction model [GDPSIQSM] V1.0, both of which integrated questionnaires and serum markers. The model's output results were compared with the results of laboratory tests and gastroscopy (including pathology), and the model's accuracy was analyzed.

The sensitivity of the gastric function prediction model was 91.16%, the specificity was 39.18%, the positive predictive value was 83.27%, the negative predictive value was 57.14%, the area under the receiver operating characteristic curve (AUC) was 0.702, and the Youden index was 0.303. The sensitivity of the gastric organic lesion prediction model was 89.55%, the specificity was 33.22%, the positive predictive value was 70.86%, the negative predictive value was 63.67%, the AUC was 0.672, and the Youden index was 0.228.

Both the gastric function prediction model and the gastric organic lesion prediction model demonstrate high sensitivity and good predictive performance, and may have potential clinical applicability.