基于多因素Logistic回归的中药治疗非哺乳期乳腺炎疗效预测模型构建与验证

Objective To explore the factors influencing the therapeutic effect of Traditional Chinese Medicine in the treatment of non-lactational mastitis(NPM), construct a prediction model for the therapeutic effect of Traditional Chinese Medicine, and verify its value. Methods Clinical data of patients with NPM who received Traditional Chinese Medicine treatment at Lishui Central Hospital and Lishui Maternal and Child Health Hospital from January 2018 to May 2024 were collected. Patients were divided into the good therapeutic effect group(recovery) and the poor therapeutic effect group(non-recovery/recurrence) based on their treatment outcomes within half a year. Univariate analysis was used to screen prognostic factors, and multivariate Logistic regression analysis was included to establish a Nomogram prediction model. Internal validation of the model was conducted using the Bootstrap method, and ROC curves were drawn. Results This study included a total of 148 cases treated with Traditional Chinese Medicine. The patients' age ranged from 23 to 61 years, with an average of(35.4±7.9) years. In the Traditional Chinese Medicine treatment group, 126 cases were recovered and 22 cases were non-recovered/recurrent. Univariate analysis showed that skin ulceration, surgical history, antibiotic use history, duration of Traditional Chinese Medicine treatment, and age were related to the therapeutic effect of Traditional Chinese Medicine(P<0.05). Multivariate Logistic regression analysis showed that skin ulceration, duration of Traditional Chinese Medicine treatment<6 months, age<35 years, and combined use of antibiotics were independent predictors of the therapeutic effect of Traditional Chinese Medicine. Internal validation by Bootstrap showed that the AUC of the modeling group was 0.826(95%CI:0.743～0.909), Hosmer-Lemeshow test(χ2=6.847, P=0.553), indicating good model calibration. Conclusion Skin ulceration, duration of Traditional Chinese Medicine treatment, age, and antibiotic use are independent factors affecting the therapeutic effect of Traditional Chinese Medicine in the treatment of NPM. The Nomogram prediction model constructed based on multivariate Logistic regression has the potential to effectively predict the therapeutic effect of Traditional Chinese Medicine in the treatment of NPM.