Single-Examination Risk Prediction of Severe Retinopathy of Prematurity

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.

Prediction of the Reactivation of Retinopathy of Prematurity After Anti-VEGF Treatment Using Machine Learning in Small Numbers

“…Some experts have proposed that a screening system based solely on vessel characteristics in the posterior pole is su cient for ROP prediction, diagnosis, progression, and therapy response. [31,[39][40][41] Gupta [39] showed that patients with ROP reactivation had a higher vascular severity score before treatment, and the ROP vascular severity score at the time of initial treatment was associated with reactivation, which suggests that retinal photographs before initial treatment may be predictive of reactivation after treatment. In our study, we used posterior fundus images before treatment to build a prediction model, achieving 83.35% accuracy, 93.33% sensitivity, and 72.91% speci city.…”

“…With the development of DL applications, more studies are trying to combine biometrical information with clinical information. Coyner [41] successfully improved the speci city of the TR-ROP prediction model by adding biometrical information; the GA + vascular severity score (VSS) is the best-performing model. The DLR-A model, added with clinical information using deep learning radiomics (DLR), obtains overall performance in a pancreatic neuroendocrine neoplasm reactivation model after the radical surgery constructed by…”

Multiomics, artificial intelligence, and precision medicine in perinatology

“…An excellent example of the use of AI is retinopathy of prematurity (ROP), which can now be detected much earlier and more accurately using these methods 37 – 39 . As an illustration ( Figure 5 ), features seen in the retina during development in preterm infants can be utilized using user defined definitions or machine learned features to provide accurate diagnosis of this disease 1 .…”

Current Management of Retinopathy of Prematurity

“…In addition, the CNN was found to be accurate in the diagnosis of ROP stages 1, 2, and 3 in fundus images from datasets in North America and Nepal [32]. When modeled, gestational age plus VSS from a single examination (at 32 to 33 weeks postmenstrual age) was found to identify all infants who developed treatment requiring ROP more than 1 month before diagnosis with moderate to high specificity [33].…”