A novel, non-invasive method for predicting cardiac illness in premature infants is provided via deep learning analysis of retinal pictures. These models have the potential to advance precision medicine in neonatal care by enabling early risk classification, better monitoring, and prompt therapies through the identification of subtle vascular and anatomical patterns. Therefore, this study evaluated whether features linked to BPD or PH in infants could be found in images acquired as part of retinopathy of prematurity (ROP) screening.
Retinal pictures from patients participating in the multi-institutional Imaging and Informatics in Retinopathy of Prematurity (i-ROP) project were analyzed using a deep learning model. From 2012 to 2020, infants at risk for ROP who underwent routine ROP screening exams were included in the analysis. Seven neonatal intensive care units provided the infants. In order to avoid the clinical diagnosis of BPD or PH, images were restricted to 34 weeks or less postmenstrual age (PMA).
This analysis included 493 newborns (mean [SD] gestational age, BPD, 25.7 [1.8] weeks; normal, 27.3 [1.8] weeks; 267 male [54.2%]). A held-out test set of 37 patients from the PH cohort and 99 patients from the BPD cohort was used to report performance. The accuracy of the multimodal model for BPD was higher than that of the demographics-only (0.72; ∆AUC, 0.1; 95% CI, −0.008 to 0.21; P =.07) or imaging-only (0.72; ∆AUC, 0.1; 95% CI, 0.04-0.16; P =.002) models.
The multimodal AUC for PH was 0.91 compared to the demographics-only 0.68 (∆AUC, 0.14; 95% CI, 0.006-0.27; P =.04) and imaging-only 0.91 (∆AUC, −0.09; 95% CI, −0.3 to 0.12; P =.40) models. When trained on photos without clinical ROP symptoms, the results remained consistent. Overall, in preterm newborns, retinal pictures acquired during ROP screening may be utilized to predict the diagnosis of BPD and PH.
Source:
Singh, P., Kumar, S., Tyagi, R., Young, B. K., Jordan, B. K., Scottoline, B., Evers, P. D., Ostmo, S., Coyner, A. S., Lin, W.-C., Gupta, A., Erdogmus, D., Chan, R. V. P., McCourt, E. A., Barry, J. S., McEvoy, C. T., Chiang, M. F., Campbell, J. P., & Kalpathy-Cramer, J. (2026). Deep learning-based prediction of cardiopulmonary disease in retinal images of premature infants. JAMA Ophthalmology. https://doi.org/10.1001/jamaophthalmol.2025.5814
Disclaimer: This website is primarily for healthcare professionals. The content here does not replace medical advice and should not be used as medical, diagnostic, endorsement, treatment, or prescription advice. Medical science evolves rapidly, and we strive to keep our information current. If you find any discrepancies, please contact us at corrections@medicaldialogues.in. Read our Correction Policy here. Nothing here should be used as a substitute for medical advice, diagnosis, or treatment. We do not endorse any healthcare advice that contradicts a physician's guidance. Use of this site is subject to our Terms of Use, Privacy Policy, and Advertisement Policy. For more details, read our Full Disclaimer here.
NOTE: Join us in combating medical misinformation. If you encounter a questionable health, medical, or medical education claim, email us at factcheck@medicaldialogues.in for evaluation.