Decision tree model to predict ovarian tumor malignancy based on clinical markers and preoperative circulating blood cells

Abstract

Abstract Background: Ovarian cancer is a serious malignant tumor that threatens women's health. And about 70% of ovarian cancers are in advanced stages when discovered. Currently, early diagnosis of ovarian cancer remains inadequate and effective treatments are lacking. Therefore, this study aims to use the decision tree method of artificial intelligence machine learning to build a model for predicting the benign and malignant degree of ovarian cancer patients. Methods: This study retrospectively analyzed 758 patients with ovarian cancer who were admitted to the gynecology department of the Second Affiliated Hospital of Xi 'an Jiaotong University from January 2018 to December 2020. The patients were diagnosed by B-ultrasound, CT or MR. The clinicopathological features and circulating blood cell indexes were recorded and analyzed. It included age, BMI, course of disease, HE4, CA125, menopausal status, general information, ROMA index before and after menopause, tumor size and location, presence or absence of ascites, red blood cell related indexes, white blood cell related indexes and platelet related indexes. Finally, the prediction model of benign and malignant ovarian tumors was constructed by CART decision tree, and the subject working curve was drawn to evaluate the predictive value of the decision tree model. Results: In this study, after statistical analysis, it was found that significant predictor variables include age, disease duration, patient general condition and menopausal status, ascites, tumor location and characteristics, HE4, CA125, ROMA index, and blood routine related indicators (except for tropism basal granulocyte percentage and absolute value). The multicollinearity between the independent variables was not obvious. In the constructed decision tree model, ROMA_after was the root node with the maximum information gain. This decision tree used indicators such as ROMA_after, CA125, PLT, Age, LY%, LY and HE4. The area under the receiver operating characteristic curve (AUC) of this model for predicting benign and malignant ovarian cancer was 0.89. Conclusion: The decision tree model was successfully constructed based on clinical indicators and preoperative circulating blood cells. Our decision tree model showed better results in predicting benign and malignant ovarian cancer than imaging indicators and biomarkers alone. This means that our model can more accurately predict benign and malignant ovarian cancer.