Deep learning of personal health, genomics, and imaging data for early detection of prostate and breast cancers
The aim of this project is to develop a multi-modality cancer risk model based on deep learning of personal health, genomics, and imaging data for early detection of prostate and breast cancers. The scientific rationale for this project is that, current screening methods for prostate and breast cancers have led to substantial over-diagnosis and over-treatment, adding no benefits to millions of cancer patients worldwide while impairing their quality of life. In this setting, the development of a multi-modality cancer risk model will help to both improve the way we stratify and detect these cancers at early stages and identify the relevant risk factors for early prevention of these diseases. This project will last 36 months. Upon completion, this project will provide an accurate, non-invasive, and cost-effective approach for prostate and breast cancer screening to avoid over-diagnosis and over-treatment, as well as advance our understanding of the correlations between these cancers and various cancer risk factors for targeted prevention.