A team of researchers at Johns Hopkins University has developed an AI-powered diagnostic tool that can detect early-stage cancers in blood samples with 97 percent accuracy across 12 different cancer types. The system, called OncoDiagnost, analyzes cell-free DNA fragments and protein biomarkers using a deep learning model trained on over 50,000 patient samples.
In a clinical trial involving 8,200 participants, OncoDiagnost correctly identified cancers at stages I and II in 97 percent of cases, compared to 52 percent for standard screening methods. The tool was particularly effective at detecting pancreatic, ovarian, and liver cancers, which are notoriously difficult to diagnose early through conventional means.
The research team has submitted the tool for FDA breakthrough device designation and expects to begin a larger Phase III clinical trial later this year. If approved, OncoDiagnost could be administered as a routine annual blood test, potentially transforming cancer screening by catching tumors years earlier than current methods allow.
