Project Grant R01CA289249

This Project Grant award from the National Cancer Institute under the Cancer Detection and Diagnosis Research program (CFDA 93.394) aims to develop a comprehensive and scalable risk prediction model, termed "PRECISE", that fuses imaging and non-imaging data to enable early detection of pancreatic cancer in asymptomatic individuals. The project, awarded $1,199,514, will run from Sep 16, 2024 to Aug 31, 2029. Key components include:

1. Developing deep learning models to segment imaging biomarkers from abdominal CT scans, applying adversarial debiasing techniques to ensure fairness across diverse patient factors and image acquisition methods. This will involve validating the models using data from Mayo Clinic, Cornell University, and UCSF.

2. Creating a graph neural network-based fusion model, PRECISE, that combines the imaging biomarkers with clinical data from electronic medical records to predict pancreatic cancer risk. The model's prognostic performance will be compared to baseline models.

3. Deploying and prospectively evaluating the PRECISE model across different geographical sites, assessing its performance by comparing predictions with patient outcomes.

The goal is to create an unbiased, effective risk prediction tool that leverages multimodal data to transform pancreatic cancer early detection, potentially enabling opportunistic screening during routine CT scans.