A Machine Learning Framework Enables Supervised Treatment Response Prediction from Tumor Transcriptomics across Cancer Types.

A machine learning framework called EXPRESSO accurately predicts treatment responses across nine cancer types using tumor transcriptomics, outperforming existing signatures in 91 cohorts.

• Why it matters: Accurate biomarkers are essential for personalized cancer treatment, but RNA transcriptomics remains underutilized due to limited data and lack of robust models, hindering precision oncology.
• What they did: The study assembled the largest transcriptomic dataset to date and developed EXPRESSO, a supervised machine learning method that integrates drug targets and biomarkers to predict responses in 5,675 patients across six therapies.
• The result: EXPRESSO successfully predicted responses and stratified progression-free survival beyond binary outcomes, demonstrating potential to improve treatment decision-making and highlighting the need for more data and mechanistic insights.