Clinical diagnosis model Construction of endometriosis based on clinical data*

Abstract

Objective Screen the relevant diagnostic indicators of endometriosis, build a diagnostic model and verify it, so as to provide a scientific basis for diagnosis and differentiation.zig. Method(s) A total of 625 patients with pathologically confirmed endometriosis were selected from December 2016 to June 2022 in Hainan Provincial people's Hospital. 308 patients with endometriosis were selected as case group and 317 patients without endometriosis as control group. There were 41 cases in the case group and 28 cases in the control group. The clinical characteristics and laboratory indexes of patients in the case group and the control group were compared: age, dysmenorrhea, progressive aggravation of symptoms, dysuria, abnormal menstruation, difficulty in sexual intercourse, low back and abdominal pain, infertility, carbohydrate antigen 125, monocyte percentage, monocyte absolute value, platelet, mean platelet volume, platelet volume distribution width, platelet volume ratio, lactate dehydrogenase, alkaline phosphatase. The independent risk factors were screened by binary Logistic regression analysis and the prediction model was constructed. Hosmer-Lemeshow was used to test the goodness of fit of the model and the subject working characteristic curve was used to judge the prediction efficiency of the model. Result(s) There were significant differences in age, dysmenorrhea, progressive aggravation of symptoms, abnormal menstruation, infertility, CA125, PCT, LDH and ALP between the two groups. The higher the CA125, the higher the risk of endometriosis, with statistical significance [OR = 1.023 (95% CI:1.016–1.029)], dysmenorrhea symptoms [OR = 3.467 (95% CI:2.052–5.859)], progressive symptoms [OR = 4.501 (95% CI:1.389–14.584)] and infertility [OR = 2.776 (95% CI:1.216–6.335)]. The higher the risk of endometriosis. The higher the LDH [OR = 0.993 (95% CI:0.987–0.999)] and the higher the ALP [OR = 0.977 (95% CI:0.962–0.991)], the lower the risk of endometriosis. The constructed model was verified by Hmurl and the result showed that P = 0.103, which suggested that the model fitted well. When the area under the model curve was 0.846 (95%CI:0.815–0.873) and the Jordan index was 0.5498, the best critical value was 0.478, the sensitivity was 69.81 and the specificity was 85.17. Conclusion(s) The model has good degree of fit and distinguishing ability, and can be used as an auxiliary means.