Column Chart Prediction Model for Ovarian Cancer Based on Serum Ovarian Tumor Related Biomarkers and Validation.

Purpose: The aim was to study the predictive model and validate serum ovarian tumor-related biomarkers for ovarian cancer histograms.

Method: We randomly selected 181 patients with ovarian tumors and 80 healthy individuals who underwent physical examinations from the hospital's medical record information system as the study participants. Clinical data and detection results of ovarian tumor-related markers such as serum carcinoembryonic antigen (CEA), carbohydrate antigen 125 (CA125), carbohydrate antigen 19-9 (CA19-9), and human epididymal protein (HE4) were collected from all study participants for analysis.

Result: Significant differences were found in serum CEA, CA125, CA19-9, and HE4 levels between healthy controls, benign ovarian tumors, and ovarian cancer (P<0.05). Dysmenorrhea (present), family history (present), age at menarche, menstrual period, number of pregnancies, natural abortion frequency, number of induced abortions, CEA, CA125, CA19-9, HE4 were all influencing factors for the incidence of ovarian cancer (P<0.05). The number of induced abortions, CEA, CA125, CA19-9, and HE4 were all independent risk factors for ovarian cancer, while the natural abortion frequency was a protective factor for ovarian cancer (P<0.05). The constructed column chart prediction model had good discrimination and prediction accuracy for ovarian cancer, good clinical utility, and higher predictive performance for ovarian cancer than traditional ROMA models.

Conclusion: The ovarian cancer column chart prediction model based on serum ovarian tumor related markers has good discrimination and prediction accuracy for ovarian cancer, with high clinical utility. Future research may need to incorporate more serum markers related to ovarian cancer to further improve the performance of predictive models.