AI Model Improves Breast Cancer Survival Prediction

Traditional survival models can estimate risk but may not fully capture the complex relationships among these variables, limiting confidence in treatment and follow-up planning. Researchers have now developed an artificial intelligence-based survival model designed to provide more individualized prognostic information and support personalized breast cancer care.

The University of Cincinnati Cancer Center developed a hybrid statistical deep learning framework for breast cancer survival prediction. The model combines statistical methods with deep learning to identify complex patterns across patient data and stratify patients who may appear clinically similar into more specific risk groups. This approach could help physicians better tailor treatment decisions and discussions about prognosis.

The model also addresses the limited availability and cost of tumor genetic testing at some hospitals. Researchers generated simulated gene expression data and integrated these molecular features with clinical information. The framework transforms clinical data into a more informative representation of patient risk, accounts for similarities among patients, and incorporates uncertainty into its survival predictions.

The team evaluated the model using a publicly available dataset containing clinical and survival information from 1,980 patients with breast cancer. In this dataset, the model achieved greater accuracy in predicting survival outcomes than the comparison models evaluated in the study. The research was published in the Journal of Statistical Theory and Practice, and the team plans to further assess the model using larger and more diverse patient datasets.

“These deep learning–based models are not meant to replace physicians. They are tools that could help physicians understand complex patient information and support their decisions. This combination of statistics, machine learning and cancer research has been exciting for me to help us improve how we understand patient outcomes,” said Wedad Alatebi, doctoral student in the Department of Biostatistics, Health Information and Data Science in the University of Cincinnati College of Medicine.

Related Links
University of Cincinnati Cancer Center