External Validation of MERCI, a machine learning algortihm for the early detection of retinal toxicity.

Chronic diseases incur a significant socio-economic cost to both patients and the health care system. Physicians must continuously monitor and modify therapies to best meet the patients’ changing requirements, often requiring regular health care visits for specialized testing.
Hydroxychloroquine (HCQ) is a disease-modifying antirheumatic drug often prescribed to patients with chronic inflammatory diseases such as rheumatoid arthritis and systemic lupus erythematosus (SLE). HCQ is generally safe but can occasionally cause sight threatening damage to the eye. It is recommended that patients treated with HCQ receive annual eye care appointments with a battery of specialized testing including visual fields, optical coherence tomography, retinal imaging and multifocal electroretinography.
We developed “The Multifocal Electroretinogram Classification Interface (MERCI)”, a machine learning approach to automate the identification of patients at risk of vision loss due to HCQ. By establishing and maintaining an open, accessible and sustainable archive of high-quality bio-specimens and rich annotated clinical data from diverse clinical populations recruited from multiple centers across Canada, this proposal will validate the performance of the MERCI algorithm and support it’s uptake as a clinical tool.

Faculty Supervisor:

Brian Ballios

Student:

Partner:

Kensington Eye Institute;Kensington Health

Discipline:

Life Sciences

Sector:

Health and Related Sciences & Technology

University:

University Health Network

Program: