Objective To compare the effects of decision tree C5.0 and Logistic regression model in predicting postpartum diastasis recti abdominis.
Methods A total of 476 puerperas with postpartum re-examination were selected as research objects. A questionnaire survey method was used to obtain general information of puerperas; an electrical stimulation therapy device was used to assess the electromyography values of the pelvic floor muscles; abdominal palpation method was used to determine the degree of diastasis recti abdominis. All the data were divided into training set and test set at a ratio of 3∶2, and the decision tree C5.0 and Logistic regression were used to establish the risk prediction models for postpartum diastasis recti abdominis; the accuracy, sensitivity, specificity, Youden index, negative predictive value, positive predictive value and area under curve (AUC) of the receiver operating characteristic (ROC) curve were used to compare the prediction performance of the models.
Results In the training set, the accuracies of decision tree C5.0 and Logistic regression model were 96.94% and 72.45% respectively, the sensitivities were 98.92% and 86.02% respectively, the specificities were 93.52% and 49.07% respectively, the positive predictive values were 96.34% and 74.42% respectively, the negative predictive values were 98.06% and 67.09% respectively, the Youden index values were 92.44% and 35.10% respectively, and the AUC values were 0.962 and 0.675 respectively; there was a significant difference in AUC between decision tree C5.0 and Logistic regression model in the training set (P < 0.05). In the test set, the accuracies of decision tree C5.0 and Logistic regression model were 81.50% and 62.43% respectively, the sensitivities were 88.35% and 82.52% respectively, the specificities were 71.43% and 32.86% respectively, the positive predictive values were 81.98% and 64.39% respectively, the negative predictive value were 80.65% and 56.10% respectively, the Youden index values were 59.78% and 15.38% respectively, and the AUC values were 0.799 and 0.577 respectively; there was a significant difference in AUC between decision tree C5.0 and Logistic regression model in the test set (P < 0.05).
Conclusion Decision tree C5.0 is better than Logistic regression model in predicting postpartum diastasis recti abdominis.
DownLoad: