Abstract
Purpose :
To evaluate the effectiveness and determine if an Artificial Intelligent (AI)/ machine learning (ML) platform can learn and recognize OCT changes in patients in patients with Wet Age Related Macular degeneration.
Methods :
This is an ongoing phase 2 analytic study of the Retina Metrics- Synthetic Analyst AI platform. 200 patients with neovascular AMD were primarily enrolled. Patients were eligible if they had received ≥3 doses of anti-VEGF therapy pro re nata (PRN) in the preceding 18 months. Changes in VA (Early Treatment Diabetic Retinopathy Study [ETDRS] letters), central foveal thickness (CFT), fluorescein leakage, and resolution of macular edema are assessed. The last-observation-carried-forward method is used to impute missing data. Safety monitoring includes monthly evaluation of ocular and systemic adverse events.
Results :
Results: Of 200 patients, changes in VA, IOP, CFT were mapped, time stamped and followed. The Retina Metrics – Synthetic Analyst platform was used to analyze, assess, interpret and potentially learn the associated clinical numeric and image findings. From baseline to date, (n=100) CFT improved by a mean (± SEM) of 39.0 ± 6.8 and 44.5 ± 7.0 mm at months 6 and 12, respectively. At study month 12, compared with baseline: in patients previously treated for <12 months (n=26) VA improved from 47.6 to 52.5 letters; in patients previously treated for 12 to 18 months (n=30) VA improved from 52.7 to 59.5 letters; and in patients previously treated for >18 months (n = 65) VA improved from 35.2 to 40.5 letters. One patient experienced a transient ischemic attack, and 1 had a retinal tear prior to month 6.
Conclusions :
Monthly Anti VEGF therapy improved VA and CFT in neovascular AMD patients who received prior PRN anti-VEGF therapy. The Synthetic Analyst AI Platform, had the ability to map and identify the both the image changes and quasi-Image changes associated within this complex patient population. Patients previously treated for >18 months with anti-VEGF therapy had worse baseline VA and final VA after Anti-VEGF therapy compared with patients previously treated for ≤18 months. With further Machine learning, the AI platform may not only be able to identify clinical changes, but it may be able to suggest a patient specific regimen aimed at improving visual acuity (VA) and anatomic outcomes in patients with neovascular age-related macular degeneration (AMD).
This is an abstract that was submitted for the 2018 ARVO Annual Meeting, held in Honolulu, Hawaii, April 29 - May 3, 2018.