Predicting cancer-specific mortality using histopathology slides

Principal Investigator

Name
Shu Jiang

Degrees
PhD

Institution
Washington University in St. Louis

Position Title
Associate professor

Email
jiang.shu@wustl.edu

About this CDAS Project

Study
PLCO (Learn more about this study)

Project ID
PLCOI-1532

Initial CDAS Request Approval
Apr 18, 2024

Title
Predicting cancer-specific mortality using histopathology slides

Summary
We have developed an AI model trained using the TCGA pan-cancer dataset to predict cancer-specific mortality. The AI algorithm intakes the digital histopathology slides under 20x or 40x magnification, and outputs a personal-level survival estimate for the next 5 or 10 years. Evaluation of the accuracy for prediction is assessed using AUC measure. We aim to utilize the PLCO pathology images for multiple cancer sites as an external validation to our pre-trained model. This will generate insights on risk stratification for patients whereby a personalized treatment and followup strategy can be generated.

Aims

1. Externally validate cancer-specific mortality using the PLCO histopathology images, report AUC measure for 5- and 10-year survival.
2. Subgroup analysis to identify cluster of patients who are at very high and very low risk of mortality

Collaborators

Dr. Graham A. Colditz (Washington University school of medicine)
Dr. Henry Zhou (Harvard TH Chan School of Public Health)