Evaluate Artera’s Deep Learning clinical-histologic system as a prognostic classifier for colorectal cancer patients

Principal Investigator

Name
Douglas Peters

Degrees
Ph.D.

Institution
Artera Inc.

Position Title
Scientific Director

Email
doug.peters@artera.ai

About this CDAS Project

Study
PLCO (Learn more about this study)

Project ID
PLCOI-2073

Initial CDAS Request Approval
Aug 24, 2026

Title
Evaluate Artera’s Deep Learning clinical-histologic system as a prognostic classifier for colorectal cancer patients

Summary
Colorectal cancer is a highly heterogeneous disease, and is the third most common cancer worldwide (PMID: 33208072). Approximately 70% of all colorectal cancer diagnoses are colon cancer. While there are several prognostic tests clinically available, there remains an unmet need for more effective risk stratification in early-stage colon cancer to better guide adjuvant treatment decisions (NCCN Colon Cancer Guidelines v4.2023). Circulating tumor DNA (ctDNA) is an emerging prognostic tool under investigation for its potential to inform treatment decisions, particularly in assessing the benefit of adjuvant chemotherapy. Although ctDNA has demonstrated prognostic value in the post-operative setting (PMID: 36646802), it may not capture the full complexity of recurrence and can be resource-intensive. To address this gap, we propose to explore whether Artera’s MMAI approach may provide prognostic value in a more accessible manner..
Artera is a company focused on developing AI approaches to personalized cancer therapy. They are led by an AI team who have previously published dozens of papers, including seminal works in Nature (PMID 28117445), Cell (PMID: 29656897), and Nature Medicine (PMID: 30617335), among many others. They utilize the Deep Learning method to develop prognostic and predictive biomarkers. Recent publications and presentations by Artera and collaborators show that their prognostic and predictive models, developed from H&E image data along with minimal clinical features, have higher AUC compared to clinical classifiers in prostate cancer. These H&E image-based methods have the advantage of increased availability compared to archival tissue, conserving scarce tissue resources, lower processing cost, and faster turn-around time.
Artera is currently rapidly expanding in other cancer indications and disease settings, such as colorectal cancer. For this project, we propose to advance the field by utilizing Artera’s AI approach to develop and validate AI-based tests that provide prognostic and predictive insights for colorectal cancer patients. This project is related to PLCOI-1961.

Aims

We propose to apply Artera’s DL methods, using baseline clinical variables, high resolution scans of available histology/pathology slides, and outcomes data from PLCO cohorts to evaluate prognostic and predictive capabilities for colorectal cancer.

Collaborators

Douglas Peter Artera Inc.
MEGAN COY Artera Inc.
Tamara Todorovic Artera Inc.
Alexandra Kraft Artera Inc.
Laura Chang Artera Inc.