< Back to all clusters
[HEALTH] · Denmark, Spain · 3 sources

ProteinTalks AI model designed to personalize breast cancer treatments

Researchers have developed ProteinTalks, an artificial intelligence tool designed to function as a 'virtual cell' to advance precision oncology for breast cancer. Published in the journal Nature, the model aims to predict how cancer cells dynamically respond to treatments by tracking protein evolution over time rather than relying on static molecular snapshots.

The study utilized a massive dataset, analyzing 18 breast cancer cell lines exposed to 63 approved anticancer drugs and 59 drug combinations. This involved over 38 million measurements across 5,585 proteins, recorded before treatment and at 6, 24, and 48-hour intervals.

ProteinTalks demonstrated the ability to predict drug efficacy, identify new therapeutic combinations, and pinpoint proteins associated with drug resistance. In testing, the model outperformed traditional machine learning methods and benchmarks based on gene activity.

Entities

Nature · ProteinTalks

Claims

What the coverage asserts, and how many sources carry each claim.