News

Indian Scientists Unveil AI System ‘OncoMark’ to Predict Cancer Cell Behaviour with High Precision

While cancer diagnosis traditionally focuses on external tumour characteristics such as size, stage and spread, these indicators often fail to fully capture a tumour’s aggressiveness. OncoMark offers a more sophisticated approach by examining the “hallmarks of cancer” — fundamental biological traits that allow cancer cells to grow uncontrollably, evade the immune system and spread across the body.

• By PTI
• New Delhi
• Dec 11, 2025

News

In a major scientific breakthrough, Indian researchers have developed a new artificial intelligence framework called OncoMark, designed to predict how cancer cells behave by analysing their underlying biological processes. The technology is expected to significantly advance the understanding of tumour progression and support future personalised cancer treatment.

While cancer diagnosis traditionally focuses on external tumour characteristics such as size, stage and spread, these indicators often fail to fully capture a tumour’s aggressiveness. OncoMark offers a more sophisticated approach by examining the “hallmarks of cancer” — fundamental biological traits that allow cancer cells to grow uncontrollably, evade the immune system and spread across the body.

These hallmarks, originally identified in landmark cancer studies and later expanded to include ten key processes, represent the molecular engines driving cancer development. Despite their importance, they are not commonly measured in current clinical assessments.

Developed by scientists from Ashoka University in Haryana and the S N Bose National Centre for Basic Sciences in Kolkata, OncoMark was trained on genetic profiles from more than 3.1 million cancer cells across 14 types of cancer. By learning how hallmark processes interact within cells, the model showed exceptional predictive capabilities. During testing, it consistently achieved accuracy rates of 96 percent or higher across multiple independent datasets.

Lead researcher Debayan Gupta noted that the AI framework can identify not only whether a cell is cancerous but also its likelihood of rapid growth, metastasis and treatment resistance. This makes OncoMark a powerful tool for reading the cellular behaviour of tumours at a molecular level.

The team envisions integrating OncoMark into clinical workflows in the future, helping oncologists gain deeper insights into a tumour’s nature and personalise therapies more effectively. Researchers also plan to expand the model to assess blood cancers and rare tumours with unique genetic and biological patterns.