Document Type

Article

Publication Date

10-9-2025

Publication Title

Translational Vision Science Technology

Abstract

Purpose: Retrospective chart reviews in ophthalmology are essential for gaining clinical insights, but they remain labor-intensive and prone to error. Despite digitization through electronic health records, extracting and interpreting lengthy, unstructured patient histories remains challenging, particularly in ophthalmology, which relies heavily on both imaging and text-based reports. We introduce OphthoACR, a Health Insurance Portability and Accountability Act-compliant artificial intelligence (AI)-powered tool for automated chart review and cohort analyses in ophthalmology. Methods: OphthoACR was applied to extract 16 variables of increasing task difficulty from the complete chart histories of 91 patients who underwent secondary intraocular lens surgery at the Columbia University Irving Medical Center from January 2020 to August 2024, for a total of 5834 unique documents. The tool integrates a fine-tuned large language model into a robust pipeline to extract and contextualize unstructured clinical data, including operative reports and imaging documents. OphthoACR's performance was compared to manual and AI-assisted chart reviews. Results: OphthoACR achieved 94% accuracy in extracting variables of interest, significantly outperforming manual review (83%). It demonstrated 97% specificity, 92% sensitivity, and a Cohen's κ of 0.70, indicating robust agreement. Average time for OphthoACR to process a patient chart was 80 seconds, a 95% reduction compared to the manual review's average of 25.2 minutes. For cohort-wide processing, the improvement was 99.9% due to parallel processing of patients' charts. Conclusions: OphthoACR significantly improves the accuracy and efficiency of ophthalmology chart reviews, offering an unprecedented automated solution to analyze large patient cohorts. Translational Relevance: OphthoACR provides end-to-end automation of retrospective chart reviews, transforming the currently labor-intensive manual process into an efficient, accurate, and scalable solution that substantially enhances clinical research.

First Page

8

PubMed ID

41065476

Volume

14

Issue

10

Creative Commons License

Creative Commons Attribution 4.0 International License
This work is licensed under a Creative Commons Attribution 4.0 International License.

Download

Share

COinS