【Research Progress】Professor Zhen Xin's Team's Latest Research Findings Indicate: An Artificial Intelligence Model Based on Laboratory Test Indicators Can Accurately Diagnose Ovarian Cancer - School of Biomedical Engineering (New Version)

**Abstract:** A groundbreaking study led by Professor Xin Zhen from the School of Biomedical Engineering at Southern Medical University, in collaboration with Professor Jihong Liu from the Cancer Center of Sun Yat-sen University, Professor Qinglei Gao from Tongji Hospital affiliated with Tongji Medical College of Huazhong University of Science and Technology, and Professor Xiao Li from the Obstetrics and Gynecology Hospital of Zhejiang University School of Medicine, has been published in the prestigious journal *Lancet Digital Health* (2023 Impact Factor: 30.8). The study introduces an innovative artificial intelligence (AI) model that leverages routine laboratory test results to accurately diagnose ovarian cancer, addressing the challenges of early detection and the lack of effective biomarkers for this disease. Ovarian cancer is the deadliest gynecological malignancy, with a 5-year survival rate of only 40%. The disease's insidious onset and lack of specific symptoms often result in late diagnosis, contributing significantly to its poor prognosis. Current diagnostic markers, such as carbohydrate antigen 125 (CA125) and human epididymis protein 4 (HE4), have limited sensitivity and specificity, making early detection and intervention difficult. The development of new, more reliable biomarkers is crucial for improving the diagnosis and survival rates of ovarian cancer patients. Routine laboratory tests, including blood counts, biochemical tests, coagulation tests, and urinalysis, are among the most common and cost-effective diagnostic tools used in clinical settings and routine health check-ups. Previous research has indicated that certain common laboratory parameters, such as albumin levels, platelet counts, and neutrophil counts, are significantly associated with ovarian cancer diagnosis and prognosis. This suggests that these routine tests could serve as potential biomarkers for the disease, enhancing early detection and improving patient outcomes. In this study, the research team collected data from over 10,000 patients across three major hospitals in China—Sun Yat-sen University Cancer Center, Tongji Hospital, and the Obstetrics and Gynecology Hospital of Zhejiang University. The dataset included 98 laboratory test results and clinical features from patients with ovarian cancer, benign gynecological conditions, and healthy individuals. Using an AI fusion approach, the researchers developed a predictive model, the MCF model, which incorporates 51 laboratory test indicators and age to diagnose ovarian cancer. The MCF model demonstrated exceptional accuracy in identifying ovarian cancer, particularly in its early stages. In internal and external validation sets, the model achieved AUC (Area Under the Curve) values of 0.949 (95% CI 0.948-0.950), 0.882 (0.880-0.885), and 0.884 (0.882-0.887), respectively. These results significantly surpass the performance of traditional biomarkers CA125, HE4, and their combination, as well as seven other previously reported AI models for ovarian cancer prediction. Notably, the MCF model maintained its predictive accuracy even when CA125 and other tumor markers were unavailable, highlighting its robustness and adaptability to real-world clinical scenarios. The MCF model has been encapsulated as an open-source tool, available on GitHub (https://github.com/xinzhen-lab/OC-prediction), allowing healthcare professionals to input routine laboratory test data and age to calculate the risk of ovarian cancer. This user-friendly tool is expected to facilitate the diagnosis of ovarian cancer in routine health check-ups and in primary healthcare settings, where experience with gynecological malignancies may be limited. The widespread availability and ease of use of this model could significantly enhance the early detection and management of ovarian cancer, potentially improving patient outcomes and reducing mortality rates. Furthermore, the study revealed that other routine laboratory tests, such as D-dimer and platelet count, also contribute significantly to the prediction of ovarian cancer. This finding suggests that the physiological processes associated with these tests may play a role in the development of ovarian cancer, opening avenues for further research into the underlying mechanisms. Professor Xin Zhen from Southern Medical University and Master's student Fangjun Huang from the same institution are the co-corresponding and co-first authors, respectively, of this study. Their work represents a significant step forward in the field of ovarian cancer diagnostics, offering a practical and effective solution to a long-standing clinical challenge. **Keywords:** ovarian cancer, early diagnosis, artificial intelligence, routine laboratory tests, biomarkers, MCF model, AUC, CA125, HE4, D-dimer, platelet count, clinical validation, open-source tool, primary healthcare, survival rates, mortality.