January 16, 2023
1 min read
Lad E. Machine and deep learning (AI) algorithms for GA identification and prediction of progression to GA from intermediate AMD: The future of dry AMD management? Presented at: Retina 2023; Jan. 15-20, 2023; Koloa, Hawaii.
Lad reports receiving consulting fees from or doing contracted research for Alexion, Allegro, Annexon, Apellis, Astellas, Boehringer Ingelheim, Galimedix, Gemini Therapeutics, Iveric Bio, Janssen, LumiThera, Nanoscope Therapeutics, Neurotech, NGM Biopharmaceuticals, Novartis, Perceive Biotherapeutics, Retrotope, Roche, and Thea Laboratories, as well as having a provisional patent for the algorithm.
KOLOA, Hawaii — An artificial intelligence-driven algorithm may help predict the progression of intermediate age-related macular degeneration to geographic atrophy, according to a speaker.
“We all understand … what a huge unmet clinical need geographic atrophy is,” Eleonora Lad, MD, PhD, said at Retina 2023. “To date, no treatment has been approved to either reverse or prevent its progression.”
There is currently an incomplete understanding of the pathogenesis and progression of nonexudative AMD, as well as what the best endpoints for clinical trials are and who suitable study candidates are, she said.
Lad and colleagues at Duke University have developed an algorithm, which has shown excellent performance characteristics for both detecting OCT-determined geographic atrophy and predicting progression of intermediate AMD to geographic atrophy on a short-term basis, she said.
The algorithm was able to automatically identify specific structural features seen on OCT that are most associated with or predictive of geographic atrophy, she said.
Further validation with additional, independent datasets is needed to help determine the algorithm’s utility for predicting vision-threatening nonexudative AMD and potentially predicting outcomes of other retinal diseases, Lad said.
“Will AI replace human experts? I don’t think so,” Lad said. “Our tools have made us stronger in the past, and we may not comprehend all the new rules our machines invent, but we will benefit from them. The biggest challenge is to keep thinking of new directions for AI to explore.”