Researchers at Oregon Health & Science University (OHSU) developed scSurvival, an open-source AI model funded by NIH, that predicts cancer patient survival using single-cell RNA sequencing (scRNA-seq) data.12
The model, published in Cancer Discovery on April 21, 2026, identifies high-risk patients and specific tumor or immune cells driving the risk, outperforming traditional bulk sequencing methods.12
scSurvival employs an attention-based multiple-instance Cox regression framework to assign survival weights to individual cells, tested on over 150 patients with melanoma and liver cancer.12
It links cell populations to outcomes, such as immunotherapy responses in melanoma, potentially transforming risk assessment and treatment.12
Key researchers include Zheng Xia (corresponding author), Faming Zhao, and Tao Ren from OHSU; endorsed by NCI Director Anthony Letai.12
Sources:
1. https://www.ainexusdaily.com/article/scsurvival-ai-cancer-survival-prediction-single-cell
2. https://www.nih.gov/news-events/news-releases/nih-funded-ai-model-predicts-cancer-survival-single-cell-tumor-data