A groundbreaking study by S N Bose National Centre for Basic Sciences in collaboration with Ashoka University has introduced OncoMark, an artificial intelligence framework that analyzes cancer at the molecular level, offering a novel approach to diagnosis and treatment. Unlike traditional staging systems that rely on tumor size and spread, OncoMark examines the hallmarks of cancer—biological programs driving tumor growth, immune evasion, metastasis, and therapy resistance.
Led by Dr. Shubhasis Haldar and Dr. Debayan Gupta, the team trained OncoMark on 3.1 million single cells across 14 cancer types, creating synthetic “pseudo-biopsies” to model hallmark-driven tumor states. The AI achieved over 99% accuracy internally and maintained 96%+ accuracy across independent cohorts. Validation on 20,000 real patient samples demonstrated its broad applicability in predicting cancer behavior.
Published in Communications Biology (Nature Publishing Group), OncoMark allows doctors to visualize hallmark activity, identify aggressive cancers that may appear less harmful under traditional staging, and select drugs targeting specific tumor processes. This framework represents a significant step toward personalized, evidence-driven cancer therapy.
