Aaron S Coyner1,2, Jimmy S Chen1, Praveer Singh3,4, Robert L Schelonka5, Brian K Jordan5, Cindy T McEvoy5, Jamie E Anderson1, R V Paul Chan6, Kemal Sonmez2, Deniz Erdogmus7, Michael F Chiang8, Jayashree Kalpathy-Cramer3,4, J Peter Campbell1. 1. Departments of Ophthalmology. 2. Medical Informatics and Clinical Epidemiology. 3. Department of Radiology, Massachusetts General Hospital and Harvard Medical School, Harvard University, Boston, Massachusetts. 4. Center for Clinical Data Science, Massachusetts General Hospital and Brigham and Women's Hospital, Boston, Massachusetts. 5. Pediatrics, Oregon Health & Science University, Portland, Oregon. 6. Department of Ophthalmology and Visual Sciences, University of Illinois Chicago, Chicago, Illinois. 7. Department of Electrical and Computer Engineering, Northeastern University, Boston, Massachusetts. 8. National Eye Institute, National Institutes of Health, Bethesda, Maryland.
Abstract
BACKGROUND AND OBJECTIVES: Retinopathy of prematurity (ROP) is a leading cause of childhood blindness. Screening and treatment reduces this risk, but requires multiple examinations of infants, most of whom will not develop severe disease. Previous work has suggested that artificial intelligence may be able to detect incident severe disease (treatment-requiring retinopathy of prematurity [TR-ROP]) before clinical diagnosis. We aimed to build a risk model that combined artificial intelligence with clinical demographics to reduce the number of examinations without missing cases of TR-ROP. METHODS: Infants undergoing routine ROP screening examinations (1579 total eyes, 190 with TR-ROP) were recruited from 8 North American study centers. A vascular severity score (VSS) was derived from retinal fundus images obtained at 32 to 33 weeks' postmenstrual age. Seven ElasticNet logistic regression models were trained on all combinations of birth weight, gestational age, and VSS. The area under the precision-recall curve was used to identify the highest-performing model. RESULTS: The gestational age + VSS model had the highest performance (mean ± SD area under the precision-recall curve: 0.35 ± 0.11). On 2 different test data sets (n = 444 and n = 132), sensitivity was 100% (positive predictive value: 28.1% and 22.6%) and specificity was 48.9% and 80.8% (negative predictive value: 100.0%). CONCLUSIONS: Using a single examination, this model identified all infants who developed TR-ROP, on average, >1 month before diagnosis with moderate to high specificity. This approach could lead to earlier identification of incident severe ROP, reducing late diagnosis and treatment while simultaneously reducing the number of ROP examinations and unnecessary physiologic stress for low-risk infants.
BACKGROUND AND OBJECTIVES: Retinopathy of prematurity (ROP) is a leading cause of childhood blindness. Screening and treatment reduces this risk, but requires multiple examinations of infants, most of whom will not develop severe disease. Previous work has suggested that artificial intelligence may be able to detect incident severe disease (treatment-requiring retinopathy of prematurity [TR-ROP]) before clinical diagnosis. We aimed to build a risk model that combined artificial intelligence with clinical demographics to reduce the number of examinations without missing cases of TR-ROP. METHODS: Infants undergoing routine ROP screening examinations (1579 total eyes, 190 with TR-ROP) were recruited from 8 North American study centers. A vascular severity score (VSS) was derived from retinal fundus images obtained at 32 to 33 weeks' postmenstrual age. Seven ElasticNet logistic regression models were trained on all combinations of birth weight, gestational age, and VSS. The area under the precision-recall curve was used to identify the highest-performing model. RESULTS: The gestational age + VSS model had the highest performance (mean ± SD area under the precision-recall curve: 0.35 ± 0.11). On 2 different test data sets (n = 444 and n = 132), sensitivity was 100% (positive predictive value: 28.1% and 22.6%) and specificity was 48.9% and 80.8% (negative predictive value: 100.0%). CONCLUSIONS: Using a single examination, this model identified all infants who developed TR-ROP, on average, >1 month before diagnosis with moderate to high specificity. This approach could lead to earlier identification of incident severe ROP, reducing late diagnosis and treatment while simultaneously reducing the number of ROP examinations and unnecessary physiologic stress for low-risk infants.
Authors: Amy K Hutchinson; Michele Melia; Michael B Yang; Deborah K VanderVeen; Lorri B Wilson; Scott R Lambert Journal: Ophthalmology Date: 2016-01-28 Impact factor: 12.079
Authors: James M Brown; J Peter Campbell; Andrew Beers; Ken Chang; Susan Ostmo; R V Paul Chan; Jennifer Dy; Deniz Erdogmus; Stratis Ioannidis; Jayashree Kalpathy-Cramer; Michael F Chiang Journal: JAMA Ophthalmol Date: 2018-07-01 Impact factor: 7.389
Authors: Miles F Greenwald; Ian D Danford; Malika Shahrawat; Susan Ostmo; James Brown; Jayashree Kalpathy-Cramer; Kacy Bradshaw; Robert Schelonka; Howard S Cohen; R V Paul Chan; Michael F Chiang; J Peter Campbell Journal: J AAPOS Date: 2020-04-11 Impact factor: 1.220
Authors: Stanford Taylor; James M Brown; Kishan Gupta; J Peter Campbell; Susan Ostmo; R V Paul Chan; Jennifer Dy; Deniz Erdogmus; Stratis Ioannidis; Sang J Kim; Jayashree Kalpathy-Cramer; Michael F Chiang Journal: JAMA Ophthalmol Date: 2019-07-03 Impact factor: 7.389
Authors: Adam Hanif; Charles Lu; Ken Chang; Praveer Singh; Aaron S Coyner; James M Brown; Susan Ostmo; Robison V Paul Chan; Daniel Rubin; Michael F Chiang; Jayashree Kalpathy-Cramer; John Peter Campbell Journal: Ophthalmol Retina Date: 2022-03-16