Personalized approach to malignant struma ovarii: Insights from a web-based machine learning tool

• Published in: International journal of gynecology and obstetrics
• Main Authors: Alshwayyat, Sakhr, Abo-Elnour, Dina Essam, Dabash, Tala Yaser, Alshwayyat, Tala Abdulsalam, Alabbasi, Marah, Alshwayyat, Mustafa, Irsheidat, Kinda Akram
• Format: Journal Article
• Language: English
• Published: United States 04.08.2024
• Subjects: ovarian neoplasms; survival analysis; machine learning; clinical decision‐making; prognosis; struma ovarii
• ISSN: 0020-7292; 1879-3479; 1879-3479
• DOI: 10.1002/ijgo.15845

Malignant struma ovarii (MSO) is a rare ovarian tumor characterized by mature thyroid tissue. The diverse symptoms and uncommon nature of MSO can create difficulties in its diagnosis and treatment. This study aimed to analyze data and use machine learning methods to understand the prognostic factors and potential management strategies for MSO. In this retrospective cohort, the Surveillance, Epidemiology, and End Results (SEER) database provided the data used for this study's analysis. To identify the prognostic variables, we conducted Cox regression analysis and constructed prognostic models using five machine learning algorithms to predict the 5-year survival. A validation method incorporating the area under the curve of the receiver operating characteristic curve was used to validate the accuracy and reliability of the machine learning models. We also investigated the role of multiple therapeutic options using the Kaplan-Meier survival analysis. The study population comprised 329 patients. Multivariate Cox regression analysis revealed that older age, unmarried status, chemotherapy, and the total number of tumors in patients were poor prognostic factors. Machine learning models revealed that the multilayer perceptron accurately predicted outcomes, followed by the random forest classifier, gradient boosting classifier, K-nearest neighbors, and logistic regression models. The factors that contributed the most were age, marital status, and the total number of tumors in the patients. The present study offers a comprehensive approach for the treatment and prognosis assessment of patients with MSO. The machine learning models we have developed serve as a practical, personalized tool to aid in clinical decision-making processes.