AI Model Predicts Breast Cancer Survival Using DNA Methylation and Gene Interactions

Scientists have created a breakthrough AI model called ARTEMIS that significantly improves breast cancer prognosis prediction by incorporating both individual genetic effects and complex gene-gene interactions. This represents a major advance over current models that typically ignore how genes work together to influence disease outcomes.

The research team analyzed DNA methylation data from nine independent breast cancer cohorts totaling over 1,600 patients. They used an innovative "3-D modeling strategy" that examines both main genetic effects and gene-gene interactions simultaneously. The model focuses on epigenetic changes—modifications that affect gene expression without altering DNA sequence—which play crucial roles in cancer development and progression.

ARTEMIS demonstrated exceptional performance across multiple validation studies. The model achieved 84.4% accuracy for 3-year survival prediction and 81.6% for 5-year prediction. Patients scoring in the top 10% had 15 times higher mortality risk compared to those in the bottom 25%. The model showed excellent calibration, meaning its predictions closely matched actual outcomes across different patient populations.

A comprehensive systematic review comparing ARTEMIS to 209 existing breast cancer prediction models revealed superior performance in three key areas: prediction accuracy, ability to work across different populations, and reliability with varying sample sizes. The researchers made ARTEMIS freely available as an online tool, enabling clinicians worldwide to input patient data and receive personalized risk assessments.

This advancement could transform breast cancer care by enabling more precise treatment decisions and improved patient counseling. The model's focus on epigenetic factors also opens new avenues for targeted therapies, as DNA methylation changes are potentially reversible through specific interventions.