An Investigation into the Efficacy of Treatments for Atopic Eruption of Pregnancy

Objective To establish a predictive model that assess the treatment effectiveness for atopic eruption of pregnancy and evaluate the efficacy of current conventional therapies. Methods A retrospective analysis was conducted on the clinical data of 653 patients diagnosed with atopic eruption of pregnancy who were treated at Anhui Maternal and Child Health Hospital between January 2009 and January 2024. A total of 367 patients were categorized into the effective group based on their symptom improvement, while 286 patients were placed in the ineffective group. Multiple regression analyses were conducted to identify suitable predictors following data collection. A predictive model for atopic eruption of pregnancy and a validation model were subsequently established. Results In Lasso regression model, lambda 0.02388 and AUC value 0.8609; in Ling regression model, lambda.min 0.01245 and AUC value 0.8629; Lambda.1se 0.02388 and AUC value 0.8540; A good value prediction model can be selected by selecting the lambda values above according to the error bar diagram and coefficient trajectory diagram. The OR value calculated by single factor analysis showed that the mild disease severity had the highest efficacy (OR=1.26). Among the disease types, E type had the highest change effectiveness (OR=3.10), which could not be included in the prediction model through multivariate analysis. The three predictors of treatment effectiveness included after data cleaning were histopathological type, allergen screening, and treatment mode. The ROC curve's area under the curve (AUC) was observed to be 0.639 with a 95% confidence interval ranging from 0.598 to 0.681. The model demonstrated a sensitivity of 81.5% and a specificity of 41.6%. The Hosmer-Lemeshow goodness of fit test affirmed the model's excellent fit (P=1.000), and no statistical significance was found between the prediction probability and the observation proportion, substantiating the model's reliable calibration capability. The congruity between the working curve and the deviation correction curve in the calibration curve, along with their proximity to the ideal curve, signifies a high level of calibration in the nomogram. The clinical decision curve shows that this model predicts a net benefit. Conclusion There is a significant multicollinearity between disease severity and disease type, and both mild severity and E-type can directly predict treatment effectiveness. The prediction model constructed using histopathological type, allergen screening, and treatment mode demonstrates good predictive value for the effectiveness of treatment of atopic eruption of pregnancy. This model can be used to evaluate treatment effectiveness and provide a reference for clinicians in conducting clinical consultations and selecting appropriate treatment methods for the right patient population.