Analysis of the Factors Influencing Postpartum Depression via Logistic Regression and Decision Tree Models

Abstract Background To study the status of postpartum depression and its main influencing factors by using logistic regression and a decision tree model and to understand the psychological characteristics of puerperae to take targeted measures to improve their mental health level. Methods A general demographic data questionnaire, the Edinburgh Postnatal Depression Scale (EPDS), the Connor-Davidson Resilience Scale (CD-RISC), and the Perceived Social Support Scale (PSSS) were used to investigate 1536 parturients who came to the Child Health Care Department of a tertiary hospital in Urumqi for physical examination. Using binary classification and a logistic regression model based on the classification of the decision tree analysis of postpartum women, the CHAID algorithm was used to compare the factors influencing PPD and the differences between the two models. Results The results of the logistic regression analysis model and decision tree model revealed that the level of resilience, degree of social support, and pregnancy complications were the influencing factors of PPD (P < 0.05), among which resilience was the most important influencing factor. Conclusion Both models have predictive value for classification, and the logistic regression model is superior to the decision tree model in predicting PPD. However, both models have advantages and disadvantages and can complement each other to make the analysis results more practical.