Epidemiologic evaluation of retinopathy of prematurity severity in a large telemedicine program in India using artificial intelligence

Ophthalmology. 2023 Apr 6:S0161-6420(23)00205-1. doi: 10.1016/j.ophtha.2023.03.026. Online ahead of print.

ABSTRACT

PURPOSE: Epidemiological changes in retinopathy of prematurity (ROP) depend on neonatal care, neonatal mortality, and the ability to carefully titrate and monitor oxygen. Herein, we evaluate whether an artificial intelligence (AI) algorithm for assessing ROP severity in babies can be utilized to evaluate changes in disease epidemiology in babies from South India over a five-year period.

DESIGN: Retrospective cohort study.

PARTICIPANTS: Babies (3093) screened for ROP at neonatal care units (NCUs) across the Aravind Eye Care System (AECS) in South India.

METHODS: Images and clinical data were collected as part of routine tele-ROP screening at the AECS in India over two time periods: August 2015-October 2017 and March 2019-December 2020. Differences in clinical diagnoses of moderate (type 2) or treatment-requiring (TR) ROP over time were evaluated against birth weight, gestational age at birth, and an AI-derived ROP vascular severity score (derived from retinal fundus images obtained at the initial tele-retinal screening exam, VSS).

MAIN OUTCOME MEASURES: Differences in the proportions of type 2 or worse ROP and TR-ROP cases, and differences in VSS between time periods.

RESULTS: Over time, the proportion [95% CI] of babies with type 2 or worse ROP and TR-ROP dropped from 60.9% [53.8%, 67.7%] to 17.1% [14.0%, 20.5%] (p < 0.001) and 16.8% [11.9%, 22.7%] to 5.1% [3.4%, 7.3%] (p < 0.001), respectively. Similarly, the median [interquartile range] VSS in the population decreased from 2.9 [1.2] to 2.4 [1.8] (p < 0.001).

CONCLUSIONS: These results suggest that AI-based assessment of ROP severity may be a useful epidemiologic tool to evaluate geographic and temporal changes in ROP incidence, severity, and risk. In South India, the use of an AI-based measure of ROP severity, when applied at the population level, is strongly suggestive of dramatic improvements in primary prevention of ROP in a short period of time, coincidental to the implementation of a large ROP tele-screening program.

PMID:37030453 | DOI:10.1016/j.ophtha.2023.03.026