The purpose of this study is to enable non-invasive early detection of ovarian cancer in high-risk populations through the establishment of a multimodal machine learning model using plasma cell-free DNA fragmentomics. Plasma cell-free DNA from early stage ovarian cancer patients and healthy individuals will be subjected to whole-genome sequencing. Five diferent feature types, including Fragment Size Coverage (FSC), Fragment Size Distribution (FSD), EnD Motif (EDM), BreakPoint Motif (BPM), and Copy Number Variation (CNV) will be assessed to generate this model.

At present, there are many problems in the detection of ovarian cancer in China, such as a large number of high-risk population, lack of effective screening and management methods, and the value of vaginal ultrasound and CA125 in early screening of ovarian cancer is limited. There is an urgent need for a more sensitive screening method for ovarian cancer in clinical practice. In a more advanced window period, a group with higher risk of disease will be screened to enter clinical diagnosis, so as to achieve early prevention and treatment of early patients and win valuable opportunities for effective prevention and treatment of ovarian cancer. Although there are some studies on early screening data of ovarian cancer at home and abroad, most of them use single detection dimension or somatic mutation combined with methylation analysis. At present, the optimization of detection technology, sample accumulation or validation of prospective clinical trials are still under way. In short, the space for early screening of ovarian cancer is vast, and liquid biopsy is non-invasive, convenient and easy to accept. It is an important technical means for early screening research of ovarian cancer, and has great potential to improve the performance of early screening of ovarian cancer. In order to further verify the application value of cfDNA-based fragmentomics in early screening of ovarian cancer and better screen the high-risk population of ovarian cancer in China, this study intends to analyze the characteristics of five cfDNA fragments based on low-depth whole-genome sequencing technology (WGS), and integrate artificial intelligence machine learning technology to establish a prediction model for early screening of ovarian cancer based on cfDNA.

Inclusion Criteria:

• Age minimum 18 years
• Patients with I-IV ovarian cancer or benign tumor confirmed by pathological examination.
• Ability to understand and the willingness to sign a written informed consent document
• Non-cancer controls are sex- and age-matched individuals without presence of any tumors or nodules or any other severe chronic diseases through systematic screening

Exclusion Criteria:

• Participants must not be pregnant or breastfeeding
• Participants must not have prior cancer histories or a second non-ovarian malignancy
• Participants must not have had any form of cancer treatment before enrollment or plasma collection, including surgery, chemotherapy, radiotherapy, targeted therapy and immunotherapy
• Participants must not present medical conditions of fever or have acute or immunological diseases that required treatment 14 days before plasma collection
• Participants who underwent organ transplant or allogenic bone marrow or hematopoietic stem cell transplantation
• Participants with clinically important abnormalities or conditions unsuitable for blood collection
• Any other disease or clinical condition of participants that the researcher believes may affect the compliance of the protocol, or affect the patient's signing of the informed consent form (ICF), which is not suitable to participate in this clinical trial.

30 patients with stage I-II ovarian cancer, 30 patients with stage III-IV ovarian cancer, 40 patients with benign ovarian cancer and 30 healthy people were enrolled