Abstract
Purpose :
To explore the understanding and adoption of artificial intelligence (AI) applications within clinical ophthalmology.
Methods :
For this exploratory pilot study, semi-structured one-on-one interviews (N=18) were conducted with ophthalmologists, ophthalmology residents, fellows, and medical professionals involved in the diagnosis and treatment of eye diseases. An interview guide informed by prior research of an interdisciplinary team and clinical expertise steered the question-asking process. Virtual interviews lasted approximately 30 minutes and were analyzed with a qualitative content analysis approach.
Results :
The majority of participants were white (56%), male (61%), and aged 25-44 years (Mean=32.3). 78% were affiliated with academia and 38% were attending physicians with 1-11 years of practice. Analysis revealed all participants believed that AI informed practice was vital in ophthalmology, with many participants describing it as the “future of the profession.” While all participants were able to discuss specific applications of AI to the field, such as diagnosing diabetic retinopathy and glaucoma, very few participants had used AI in their own practice. A balance between the ‘computer and the clinician’ was noted. Identified disadvantages of an AI-integrated practice include: machine learning techniques can be subject to human data collection bias, a lack of ‘big data’ to help inform AI models, lack of buy-in from the profession, and a perceived lack of access to AI in rural areas. Participants envision integrating AI into future practice as a tool to help guide decision making and to save time. No substantial differences or patterns in responses were noted across participant demographics.
Conclusions :
Participants exhibited foundational knowledge of AI in clinical ophthalmology practice and how AI can be leveraged in their future work. However barriers to adoption and integration of AI to practice were widely noted. Looking forward, additional training on specific AI driven models to increase clinician knowledge and adoption are recommended for actualization of AI improvements in clinical practice.
This abstract was presented at the 2022 ARVO Annual Meeting, held in Denver, CO, May 1-4, 2022, and virtually.