A machine learning framework combining cfDNA fragmentomics and serum biomarkers for early ovarian cancer detection
A combined blood test merging DNA fragments and protein markers detects ovarian cancer with 97% accuracy, potentially improving screening and pre-surgery planning.
This study integrated low-coverage whole-genome sequencing-derived cfDNA features (copy number variation, fragment size, Neomer features) with serum biomarkers CA125 and HE4 in a stacked machine-learning model for ovarian cancer detection across training, validation, and external cohorts (N=195 total). The combined model achieved AUC 0.968 in validation and maintained strong performance for early-stage disease (AUC 0.938, FIGO I/II), suggesting clinical potential for pre-operative risk stratification and population screening.
What the study was
- Study design
- Prospective cohort with independent and external validation
- Population
- Ovarian cancer patients, benign ovarian disease, healthy controls; multi-site China cohort
- Sample size
- 195
- Category
- Early Detection
- Maturity
- Validated
- Journal
- Cell Communication and Signaling
Why it surfaced
Novel integrated cfDNA fragmentomics + serum biomarker ML model for ovarian cancer with high diagnostic accuracy across independent and external validation cohorts; strong performance for early-stage disease addresses a major unmet clinical need.
A plain-language summary of published research — not medical advice. Talk to a clinician about your own care.