Artificial intelligence and patient reported outcomes in ophthalmology

Abstract

Artificial intelligence (AI) has the potential to revolutionise the delivery of ophthalmic healthcare worldwide. Whether AI is making a meaningful difference or is acceptable for patients, however, remains unclear. Patient reported outcomes (PROs) allow researchers to answer these questions and smooth the path to clinical deployment. This review aims to investigate how PROs are being applied to the development and evaluation of ophthalmic AI technology and explore any underlying reasons why PROs may be currently underutilised. A systematic search of electronic databases for studies and clinical trials applying PROs in the development and evaluation of ophthalmic AI was conducted from date of inception to February 2025. Fifty articles applied a PRO to ophthalmic AI, from which 14 interventional studies and 24 unique validated PROs were identified. There was a rapid year-on-year increase in PRO utilisation beginning in 2020 until 2024. PROs were concentrated in economically advanced countries, were generic (58%) rather than disease-specific (40%), and most often were used as evaluator metrics (50%), or input (44%) for the AI model. Few articles investigated consumer-ready technologies (12%). Low research priority, the nascent state of AI in ophthalmology, and lack of high quality accumulated PROs data were identified as possible barriers to realising the full potential of PROs in ophthalmic AI. Investment into the development of robust validated PROs and the inclusion of PROs in routine data collection may catalyse the development of AI technologies capable of making the greatest meaningful difference to patients’ quality of life.

Ähnliche Arbeiten