A unified vision-language model for precision oncology and biomarker prediction in neuroblastoma.
AI reading routine tumor slides now predicts genetic markers and survival for childhood neuroblastoma, removing barriers to precision care.
NEVA is a vision-language AI that processes routine H&E pathology slides from neuroblastoma patients and predicts subtype, risk group, NMYC amplification status, and survival outcomes at accuracy exceeding ten existing foundation models across 1,238 patients from multiple institutions. The model addresses the critical gap in molecular profiling accessibility in paediatric oncology by delivering actionable genomic predictions from histology alone.
What the study was
- Study design
- multicenter_retrospective_cohort
- Category
- precision_oncology
- Maturity
- Potentially Practice-Changing
- Journal
- Nat Commun
Why it surfaced
Neuroblastoma is the leading cause of paediatric cancer mortality with major management challenges; NEVA addresses the high unmet need for accessible molecular stratification; multi-institutional 1,238-patient cohort with 11 clinically meaningful endpoints, Nature Communications.
A plain-language summary of published research — not medical advice. Talk to a clinician about your own care.