Artificial intelligence tool for rapid, automated electronic health record review

This technology is an AI-powered, automated solution for rapid, accurate, and high-quality data extraction from electronic health records.

Unmet Need: Efficient and reliable electronic clinical record and chart reviews

Researchers often rely on manual chart abstraction from electronic health records (EHRs), which is time-consuming, prone to error, and limited in scalability. Most valuable clinical data is stored in unstructured formats, like operative notes and imaging reports, making comprehensive data extraction challenging. These barriers restrict the ability to conduct large-scale, reproducible studies and hinder the discovery of insights that can improve patient care and outcomes.

The Technology: OphthoACR, an advanced automated AI electronic health record review tool

This technology, termed OphthoACR, is an AI-powered platform that automates the extraction of structured variables from complex, unstructured electronic health record (her) documentation, processing vast volumes of operative notes and imaging reports in seconds. By automating this workflow, OphthoACR overcomes the traditionally slow and error-prone process of EHR reviews, enabling rapid, reproducible cohort analysis and high-throughput clinical research.

OphthoACR achieved 94% accuracy in extracting variables of interest, significantly outperforming manual review (83%).

Applications:

  • Automated retrospective chart review for large-scale ophthalmology research
  • Identification of eligible patient cohorts for clinical trials
  • Extraction of surgical metrics and outcomes from operative notes
  • Development and maintenance of disease registries using parsed electronic health record (EHR) data
  • Support of real-time patient identification for tailored interventions
  • Large-scale analysis of imaging and narrative reports
  • Population health monitoring using aggregated structured data
  • Assembly of patient cohorts for accelerating observational studies

Advantages:

  • Fine-tuned large language model trained specifically on ophthalmic clinical documentation
  • Enables precise extraction of structured variables from unstructured EHR sources
  • Integrates with existing EHR systems
  • 95% faster patient chart processing than manual review
  • Reduced clinician documentation burden through automated data abstraction

Lead Inventor:

Leejee Han Suh, M.D.

Related Publications:

Tech Ventures Reference: