AI tool predicts cancer spread with high accuracy

A newly developed artificial intelligence system may help doctors identify which cancers are likely to spread, opening the door to earlier and more targeted treatment.
 

Researchers at the University of Geneva have created a model that analyses gene activity in tumour cells to estimate the risk of metastasis  the process by which cancer spreads to other parts of the body.
 

The tool, called Mangrove Gene Signatures (MangroveGS), was developed after scientists identified specific genetic patterns linked to the ability of cancer cells to move and form new tumours. 

Their findings were published in the journal Cell Reports.
 

Lead researcher Professor Ariel Ruiz i Altaba said cancer should be understood as a distorted form of biological development rather than random cell behaviour. According to him, certain genetic changes can reactivate early developmental processes, driving tumour growth and spread.
 

To study this, researchers examined colon cancer cells, tracking how they behave in laboratory settings and in animal models. 

They found that metastasis is influenced not only by individual cells but also by how groups of cancer cells interact.
 

By integrating these genetic patterns into the AI system, the team achieved nearly 80 percent accuracy in predicting cancer spread and recurrence in colon cancer patients. 

The same approach also showed promise in assessing risks in other cancers, including breast, lung and stomach cancer.
 

Scientists say the tool could help doctors tailor treatment plans more effectively reducing unnecessary therapy for low-risk patients while ensuring closer monitoring for those at higher risk.
 

The system can be applied to tumour samples collected in hospitals, where it analyses RNA data to generate a risk score that can be shared with clinicians for decision-making.