Related Experiment Video
Updated: Aug 27, 2025
05:40
Using an Automated Hirschberg Test App to Evaluate Ocular Alignment
Published on: March 24, 2020
12.4K
A Framework for Automating Psychiatric Distress Screening in Ophthalmology Clinics Using an EHR-Derived AI Algorithm
Samuel I Berchuck1, Alessandro A Jammal2, David Page3
1Department of Statistical Science, Duke University, Durham, NC, USA.
Translational Vision Science & Technology
|September 30, 2022
Summary
An artificial intelligence (AI) algorithm can passively monitor psychiatric distress in ophthalmology patients using electronic health record (EHR) data. This approach may enhance patient outcomes when combined with timely interventions.
Area of Science:
- Ophthalmology
- Psychiatry
- Artificial Intelligence
- Health Informatics
Background:
- Psychosocial factors significantly impact ophthalmic patient morbidity and mortality.
- Early psychiatric screening is crucial for timely intervention but is resource-intensive.
- Automated screening methods are needed to improve efficiency and accessibility.
Purpose of the Study:
- To develop and validate an artificial intelligence (AI) algorithm for automated psychiatric distress screening in ophthalmology patients.
- To leverage electronic health record (EHR) data for passive monitoring of mental health in this population.
Main Methods:
- A retrospective analysis of the Duke Ophthalmic Registry (EHR database) was performed.
- An AI algorithm was developed to identify psychiatric distress using computable phenotypes and risk factors from EHR data.
- Model performance was assessed using ROC and PR AUC, with variable importance analyzed via odds ratios.
Main Results:
- The study analyzed 358,135 encounters from 40,326 patients.
- The AI algorithm achieved an ROC AUC of 0.91 and a PR AUC of 0.55.
- Key predictors included existing distress, chemotherapy, esophageal disorders, central pain syndrome, and headaches.
Conclusions:
- An AI algorithm trained on EHR data can effectively monitor psychiatric distress in ophthalmology patients.
- This automated approach holds potential for improving health outcomes when integrated with treatment programs.
