What question did this study set out to answer?

To assess the accuracy of an AI system in automating colorectal cancer registry data abstraction from electronic medical records.

May 30, 2026Journal of Clinical Oncology0 citations

Automated Abstraction of Colorectal Cancer Registry Data Using AI

Automated abstraction of colorectal cancer registry data using AI: Accuracy and implementation insights.

View Full Paper

Ask AI

Bookmark

Authors

Randomized trial evaluates AI for improving accuracy in colorectal cancer registry data abstraction, suggesting effective automation methods.

To assess the accuracy of an AI system in automating colorectal cancer registry data abstraction from electronic medical records.
Evaluated Synapsis AI for extracting 33 colorectal cancer registry fields from EMR.
Included adult patients treated at Cleveland Clinic with specific ICD-10 codes, capped at 200 cases.
Compared AI outputs with manually abstracted tumor registry data.
Overall accuracy of 94.78% across all fields, with question-level accuracy of 100% for several key metrics.
Lowest accuracy was 28% for residual colorectal tumor presence due to ambiguity in field definitions.
Accuracy exceeded 72% across all remaining fields, highlighting areas needing improved data access and field alignment.

Lindberg et al. (2026) studied this question.

View Full Paper

Ask AI

Bookmark

May 30, 2026Journal of Clinical Oncology0 citations

Automated abstraction of colorectal cancer registry data using AI: Accuracy and implementation insights.

View Full Paper

Ask AI

Bookmark

Authors

Randomized trial evaluates AI for improving accuracy in colorectal cancer registry data abstraction, suggesting effective automation methods.

To assess the accuracy of an AI system in automating colorectal cancer registry data abstraction from electronic medical records.
Evaluated Synapsis AI for extracting 33 colorectal cancer registry fields from EMR.
Included adult patients treated at Cleveland Clinic with specific ICD-10 codes, capped at 200 cases.
Compared AI outputs with manually abstracted tumor registry data.
Overall accuracy of 94.78% across all fields, with question-level accuracy of 100% for several key metrics.
Lowest accuracy was 28% for residual colorectal tumor presence due to ambiguity in field definitions.
Accuracy exceeded 72% across all remaining fields, highlighting areas needing improved data access and field alignment.

Lindberg et al. (2026) studied this question.

View Full Paper

Ask AI

Bookmark