Author Block: Damien Pike, Stuti Tanya, Christopher S. Jackman
Author Disclosure Block: D. Pike: None. S. Tanya: None. C.S. Jackman: None.

Abstract Body:

Purpose: With Canada’s aging population the demands on ophthalmologists to provide timely and high-quality vision care are steadily increasing. To address this, the clinical and research community has harnessed artificial intelligence (AI) and machine learning (ML) to improve technological advancements and increase the efficiency with which we deliver care to patients in ophthalmology clinics. In addition to making clinical care more efficient, we believe that using sophisticated AI-based software to improve how primary care practitioners refer patients to ophthalmology services holds significant value and can play an important role to help meet the needs of our aging population. Here, our objective is to develop semi-automated software for ophthalmology referrals and assess its feasibility and efficacy in community-based ophthalmology practices.

Study Design: The design of this study was subdivided into two segments 1) to develop a novel decision-tree based software platform for ophthalmology referrals and 2) to use quantitative measurements of the software’s performance to test its efficacy in a comprehensive ophthalmology practice.

Methods: The software algorithm was initially written and developed in MATLAB R2016b using objective C++ coding language. The referral algorithm was built to prompt the referring practitioner with questions about the ocular complaint which, based on the response, would help the practitioner with next steps in the referral process.

Results: A beta version of the referral software is currently being tested internally at the Jackman Eye Institute in St. John’s, Newfoundland. Preliminary data shows that the standardized referral platform helps to guide primary care practitioners with referral decisions insofar that it provides a tentative diagnosis and timeline for the patient.

Conclusions: We are developing a software platform to improve the pipeline for ophthalmology referrals from primary care practitioners. Future work will focus on implementing convolutional neural network based ML to train the referral software to help triage the consults from referring practitioners.