Construction and Evaluation of Gastric Cancer Risk Prediction Model

To predict the probability of gastric cancer and precancerous lesions in patients and construct a risk model for gastric cancer based on related risk factors. 189 patients with gastric cancer or precancerous lesions diagnosed by pathological examination in our hospital from October 2016 to May 2021 were selected as the experimental group and 203 patients with chronic gastritis during the same period as the control group. Collect the clinical data of the two groups of patients use single factor analysis to screen and then perform multi-factor logistic regression analysis. To predict the effect, the Hosmer-Lemeshow goodness-of-fit test evaluates its degree of fit. Univariate analysis showed body mass index, family history of gastric cancer, Helicobacter pylori infection, serum pepsinogen I/II ratio, gastrin 17, fruit and vegetable intake, age of recipients, physical activity and radiation work history in patients with gastric cancer and precancerous lesions. All were higher than the control group (p<0.05). Logistic regression analysis showed that Helicobacter pylori infection, serum pepsinogen I/II ratio and gastrin 17 were independent risk factors for gastric cancer. The area under curve of the prediction model constructed based on these three indicators is 0.924 (95 % CI: 0.891 similar to 0.993) and its sensitivity and specificity are 82.41 % and 94.93 %, respectively. The Hosmer-Lemeshow goodness-of-fit test result of the prediction model showed (Hosmer-Lemeshow c(2)=13.301, p=0.358), indicating that the prediction model has a good fit. Patients with Helicobacter pylori infection, serum pepsinogen I/II ratio and gastrin 17 are independent influencing factors. The prediction model established has good efficacy in predicting the risk of gastric cancer and precancerous lesions and can be used as a clinical guide for the prevention of gastric cancer.