UniCure: A multi-modal model for predicting personalized cancer therapy response.

UniCure accurately predicts personalized cancer drug responses across diverse cell and tissue contexts using 1.9 million transcriptomic profiles.

• Why it matters: Predicting drug efficacy in individual patients remains difficult because models trained on cell lines often do not reflect patient-specific biology, limiting personalized therapy development.
• What they did: The authors developed UniCure, a multi-modal model combining biological and chemical foundation models, trained on extensive transcriptomic data, and fine-tuned on 345 patient-derived profiles, validated on 396 clinical profiles.
• The result: UniCure enables effective patient-level prediction, supports response-based stratification, and has been validated in cell line and patient-derived models, offering a practical framework for personalized cancer therapy.