A scalable deep-learning framework for cancer detection using cell-free DNA shallow whole-genome sequencing.
An AI tool detects multiple cancer types from simple blood tests, finding 21-31% of early-stage cancers reliably at lower cost than current methods.
UNITE is a tumor-naive AI framework that detects cancer across 26 types from shallow cfDNA whole-genome sequencing at 0.1x depth. In a 2063-sample validation cohort, it achieved 21-31% sensitivity for stage I-II cancers at 95% specificity, representing a scalable, cost-effective approach to multi-cancer early detection.
What the study was
- Study design
- multi-cohort machine learning development and validation
- Population
- Multi-cancer cohort, 2063 samples, 26 cancer types
- Sample size
- 2063
- Category
- Early Detection
- Maturity
- Validated
- Journal
- Science advances
Why it surfaced
Novel tumor-naive multifeature cfDNA AI (UNITE) from Cambridge/CRUK; large validation cohort (2063 samples, 26 cancer types); directly relevant to T3 and T4 watchlists; high methodological novelty.
A plain-language summary of published research — not medical advice. Talk to a clinician about your own care.