Indian researchers have developed OncoMark, an artificial intelligence–based framework capable of analysing the internal workings of cancer cells to predict tumour behaviour, potentially transforming personalised cancer treatment. Described by its creators as a way of “reading the mind of cancer,” OncoMark focuses on the hallmarks of cancer—core biological processes that normal cells manipulate to become malignant, such as evading the immune system, disabling growth suppressors, resisting cell death, and promoting metastasis. Teams at Ashoka University, Sonipat, Haryana, and the S N Bose National Centre for Basic Sciences, Kolkata, developed the framework and published the findings in Communications Biology in October 2025.
Limitations of Conventional Cancer Diagnosis
Traditional diagnostic methods primarily evaluate tumours externally, assessing size, location, and spread (staging) rather than the underlying molecular mechanisms driving cancer. Lead researcher Debayan Gupta explained, “Stages of cancer are not reliable measures of how well a patient is doing. Cancer cells manipulate their death timers and growth controls to divide indefinitely while evading immune attack.” OncoMark addresses this gap by examining the internal hallmark activities of cancer cells, offering a biology-driven insight into tumour behaviour.
Training OncoMark with Millions of Cells
The AI framework was trained on genetic data from 3.1 million cancer cells across 14 cancer types. By learning how hallmark processes such as immune evasion, proliferation, and metastasis interact, OncoMark can predict tumour growth and treatment resistance.
When validated on five independent datasets, the model achieved accuracy scores exceeding 96%, estimating both the likelihood of a cell becoming cancerous and the aggressiveness of existing tumours.
Detailed Molecular Profiles for Personalized Treatment
OncoMark calculates the probability of each hallmark’s activity to generate a comprehensive molecular profile for every tumour. “This can give doctors much deeper insight into internal cancer processes,” Gupta noted, emphasizing that integrating OncoMark into clinical workflows could help oncologists make informed, patient-specific treatment decisions at the bedside.
Expanding to Rare and Blood Cancers
As reported by indiamedtoday.com, the research team plans to extend OncoMark to blood cancers and rarer malignancies, which often behave differently from solid tumours and are challenging to characterise with conventional tools. If validated in clinical settings, OncoMark could complement traditional staging, providing oncologists with a deeper, biology-driven understanding of cancer behaviour and response, paving the way for tailored treatment strategies.