Search

Objective: This study develops and evaluates multimodal machine learning models for differentiating bacterial and fungal keratitis using a prospective representative dataset from South India. Design: Machine learning classifier training and validation study. Participants: Five hundred ninety-nine subjects diagnosed with acute infectious keratitis at Aravind Eye Hospital in Madurai, India. Methods: We developed and compared 3 prediction models to distinguish bacterial and fungal keratitis using a prospective, consecutively-collected, representative dataset gathered over a full calendar year (the MADURAI dataset). These models included a clinical data model, a computer vision model using the EfficientNet architecture, and a multimodal model combining both imaging and clinical data. We partitioned the MADURAI dataset into 70% train/validation and 30% test sets. Model training was performed with fivefold cross-validation. We also compared the performance of the MADURAI-trained computer vision model against a model with identical architecture but trained on a preexisting dataset collated from multiple prior bacterial and fungal keratitis randomized clinical trials (RCTs) (the RCT-trained computer vision model). Main Outcome Measures: The primary evaluation metric was the area under the precision-recall curve (AUPRC). Secondary metrics included area under the receiver operating characteristic curve (AUROC), accuracy, and F1 score. Results: The MADURAI-trained computer vision model outperformed the clinical data model and the RCT-trained computer vision model on the hold-out test set, with an AUPRC 0.94 (95% confidence interval: 0.92–0.96), AUROC 0.81 (0.76–0.85), accuracy 77%, and F1 score 0.85. The multimodal model did not substantially improve performance compared with the computer vision model. Conclusions: The best-performing machine learning classifier for infectious keratitis was a computer vision model trained using the MADURAI dataset. These findings suggest that image-based deep learning could significantly enhance diagnostic capabilities for infectious keratitis and emphasize the importance of using prospective, consecutively-collected, representative data for machine learning model training and evaluation. Financial Disclosure(s): Proprietary or commercial disclosure may be found in the Footnotes and Disclosures at the end of this article.

Purpose: Retinopathy of prematurity (ROP) stage is defined by the visual appearance of the vascular-avascular border, which reflects a spectrum of pathologic neurovascular tissue (NVT). Previous work demonstrated that the thickness of the ridge lesion, measured using OCT, corresponds to higher clinical diagnosis of stage. This study evaluates whether the volume of anomalous NVT (ANVTV), defined as abnormal tissue protruding from the regular contour of the retina, can be measured automatically using deep learning to develop quantitative OCT-based biomarkers in ROP. Design: Single-center retrospective case series. Participants: Thirty-three infants with ROP in the Oregon Health & Science University neonatal intensive care unit. Methods: OCT B-scans were collected using an investigational ultrawidefield OCT. The ANVTV was manually segmented. A set of 3347 B-scans and corresponding manual segmentations from 12 volumes from 6 patients were used to train an automated segmentation tool using a U-Net. An additional held-out test data set of 60 B-scans from 6 infants was used to evaluate model performance. The Dice–Sorensen coefficient (DSC) comparing manual and automated segmentation of ANVTV was calculated. Scans from 21 additional infants were used for clinical evaluation of ANVTV using the visit in which they had developed their peak stage of ROP. Each infant had every B-scan in a volume automatically segmented for ANVTV (total number of segmented voxels within the 60° temporal to the optic disc). The ANVTV was compared between infants with stage 1 to 3 ROP using a Kruskal–Wallis test and tracked over time in all infants with stage 3 ROP. Main Outcome Measurements: Cross sectional and longitudinal association between ANVTV and stages 1 to 3 ROP. Results: Comparing automated and manual segmentation of ANVTV achieved a DSC of 0.61 ± 0.13. Using the U-Net, ANVTV was associated with higher disease stage both cross sectionally and longitudinally. Median ANVTV significantly increased as ROP stage worsened from 1 (0, [interquartile range: 0–0] kilovoxels) to 2 (170.1 [interquartile range: 104.2–183.6] kilovoxels) to 3 (421.4 [interquartile range: 312.3–1110.8] kilovoxels; P < 0.001). Conclusions: Automated OCT-based measurement of ANVTV was associated with clinical disease stage in ROP, both cross sectionally and longitudinally. Ultrawidefield-OCT may facilitate more objective screening, diagnosis, and monitoring in the future. Financial Disclosure(s): Proprietary or commercial disclosure may be found in the Footnotes and Disclosures at the end of this article.

Summary: Background: It is uncertain which biological features underpin the response of rectal cancer (RC) to radiotherapy. No biomarker is currently in clinical use to select patients for treatment modifications. Methods: We identified two cohorts of patients (total N = 249) with RC treated with neoadjuvant radiotherapy (45Gy/25) plus fluoropyrimidine. This discovery set included 57 cases with pathological complete response (pCR) to chemoradiotherapy (23%). Pre-treatment cancer biopsies were assessed using transcriptome-wide mRNA expression and targeted DNA sequencing for copy number and driver mutations. Biological candidate and machine learning (ML) approaches were used to identify predictors of pCR to radiotherapy independent of tumour stage. Findings were assessed in 107 cases from an independent validation set (GSE87211). Findings: Three gene expression sets showed significant independent associations with pCR: Fibroblast-TGFβ Response Signature (F-TBRS) with radioresistance; and cytotoxic lymphocyte (CL) expression signature and consensus molecular subtype CMS1 with radiosensitivity. These associations were replicated in the validation cohort. In parallel, a gradient boosting machine model comprising the expression of 33 genes generated in the discovery cohort showed high performance in GSE87211 with 90% sensitivity, 86% specificity. Biological and ML signatures indicated similar mechanisms underlying radiation response, and showed better AUC and p-values than published transcriptomic signatures of radiation response in RC. Interpretation: RCs responding completely to chemoradiotherapy (CRT) have biological characteristics of immune response and absence of immune inhibitory TGFβ signalling. These tumours may be identified with a potential biomarker based on a 33 gene expression signature. This could help select patients likely to respond to treatment with a primary radiotherapy approach as for anal cancer. Conversely, those with predicted radioresistance may be candidates for clinical trials evaluating addition of immune-oncology agents and stromal TGFβ signalling inhibition. Funding: The Stratification in Colorectal Cancer Consortium (S:CORT) was funded by the Medical Research Council and Cancer Research UK (MR/M016587/1).

Purpose: The murine oxygen-induced retinopathy (OIR) model is one of the most widely used animal models of ischemic retinopathy, mimicking hallmark pathophysiology of initial vaso-obliteration (VO) resulting in ischemia that drives neovascularization (NV). In addition to NV and VO, human ischemic retinopathies, including retinopathy of prematurity (ROP), are characterized by increased vascular tortuosity. Vascular tortuosity is an indicator of disease severity, need to treat, and treatment response in ROP. Current literature investigating novel therapeutics in the OIR model often report their effects on NV and VO, and measurements of vascular tortuosity are less commonly performed. No standardized quantification of vascular tortuosity exists to date despite this metric’s relevance to human disease. This proof-of-concept study aimed to apply a previously published semi-automated computer-based image analysis approach (iROP-Assist) to develop a new tool to quantify vascular tortuosity in mouse models. Design: Experimental study. Subjects: C57BL/6J mice subjected to the OIR model. Methods: In a pilot study, vasculature was manually segmented on flat-mount images of OIR and normoxic (NOX) mice retinas and segmentations were analyzed with iROP-Assist to quantify vascular tortuosity metrics. In a large cohort of age-matched (postnatal day 12 [P12], P17, P25) NOX and OIR mice retinas, NV, VO, and vascular tortuosity were quantified and compared. In a third experiment, vascular tortuosity in OIR mice retinas was quantified on P17 following intravitreal injection with anti-VEGF (aflibercept) or Immunoglobulin G isotype control on P12. Main Outcome Measures: Vascular tortuosity. Results: Cumulative tortuosity index was the best metric produced by iROP-Assist for discriminating between OIR mice and NOX controls. Increased vascular tortuosity correlated with disease activity in OIR. Treatment of OIR mice with aflibercept rescued vascular tortuosity. Conclusions: Vascular tortuosity is a quantifiable feature of the OIR model that correlates with disease severity and may be quickly and accurately quantified using the iROP-Assist algorithm. Financial Disclosure(s): Proprietary or commercial disclosure may be found in the Footnotes and Disclosures at the end of this article.

Purpose: Retinopathy of prematurity (ROP) is one of the leading causes of blindness in children. Although the role of oxygen in the pathophysiology of ROP is well established, a precise understanding of the dynamic relationship between oxygen exposure ROP incidence and severity is lacking. The purpose of this study was to evaluate the correlation between time-dependent oxygen variables and the onset of ROP. Design: Retrospective cohort study. Participants: Two hundred thirty infants who were born at a single academic center and met the inclusion criteria were included. Infants are mainly born between January 2011 and October 2022. Methods: Patient data were extracted from electronic health records (EHRs), with sufficient time-dependent oxygen data. Clinical outcomes for ROP were recorded as none/mild or moderate/severe (defined as type II or worse). Mixed-effects linear models were used to compare the 2 groups in terms of dynamic oxygen variables, such as daily average and the coefficient of variation (COV) fraction of inspired oxygen (FiO2). Support vector machine (SVM) and long-short-term memory (LSTM)-based multimodal models were trained with fivefold cross-validation to predict which infants would develop moderate/severe ROP. Gestational age (GA), birth weight, and time-dependent oxygen variables were used to develop predictive models. Main Outcome Measures: Model cross-validation performance was evaluated by computing the mean area under the receiver operating characteristic (AUROC) curve, precision, recall, and F1 score. Results: We found that both daily average and COV of FiO2 were associated with more severe ROP (adjusted P

Objective: To develop a generative adversarial network (GAN) to segment major blood vessels from retinal flat-mount images from oxygen-induced retinopathy (OIR) and demonstrate the utility of these GAN-generated vessel segmentations in quantifying vascular tortuosity. Design: Development and validation of GAN. Subjects: Three datasets containing 1084, 50, and 20 flat-mount mice retina images with various stains used and ages at sacrifice acquired from previously published manuscripts. Methods: Four graders manually segmented major blood vessels from flat-mount images of retinas from OIR mice. Pix2Pix, a high-resolution GAN, was trained on 984 pairs of raw flat-mount images and manual vessel segmentations and then tested on 100 and 50 image pairs from a held-out and external test set, respectively. GAN-generated and manual vessel segmentations were then used as an input into a previously published algorithm (iROP-Assist) to generate a vascular cumulative tortuosity index (CTI) for 20 image pairs containing mouse eyes treated with aflibercept versus control. Main Outcome Measures: Mean dice coefficients were used to compare segmentation accuracy between the GAN-generated and manually annotated segmentation maps. For the image pairs treated with aflibercept versus control, mean CTIs were also calculated for both GAN-generated and manual vessel maps. Statistical significance was evaluated using Wilcoxon signed-rank tests (P ≤ 0.05 threshold for significance). Results: The dice coefficient for the GAN-generated versus manual vessel segmentations was 0.75 ± 0.27 and 0.77 ± 0.17 for the held-out test set and external test set, respectively. The mean CTI generated from the GAN-generated and manual vessel segmentations was 1.12 ± 0.07 versus 1.03 ± 0.02 (P = 0.003) and 1.06 ± 0.04 versus 1.01 ± 0.01 (P < 0.001), respectively, for eyes treated with aflibercept versus control, demonstrating that vascular tortuosity was rescued by aflibercept when quantified by GAN-generated and manual vessel segmentations. Conclusions: GANs can be used to accurately generate vessel map segmentations from flat-mount images. These vessel maps may be used to evaluate novel metrics of vascular tortuosity in OIR, such as CTI, and have the potential to accelerate research in treatments for ischemic retinopathies. Financial Disclosure(s): The author(s) have no proprietary or commercial interest in any materials discussed in this article.

Objective: To investigate the impact of corneal photograph quality on convolutional neural network (CNN) predictions. Design: A CNN trained to classify bacterial and fungal keratitis was evaluated using photographs of ulcers labeled according to 5 corneal image quality parameters: eccentric gaze direction, abnormal eyelid position, over/under-exposure, inadequate focus, and malpositioned light reflection. Participants: All eligible subjects with culture and stain-proven bacterial and/or fungal ulcers presenting to Aravind Eye Hospital in Madurai, India, between January 1, 2021 and December 31, 2021. Methods: Convolutional neural network classification performance was compared for each quality parameter, and gradient class activation heatmaps were generated to visualize regions of highest influence on CNN predictions. Main Outcome Measures: Area under the receiver operating characteristic and precision recall curves were calculated to quantify model performance. Bootstrapped confidence intervals were used for statistical comparisons. Logistic loss was calculated to measure individual prediction accuracy. Results: Individual presence of either light reflection or eyelids obscuring the corneal surface was associated with significantly higher CNN performance. No other quality parameter significantly influenced CNN performance. Qualitative review of gradient class activation heatmaps generally revealed the infiltrate as having the highest diagnostic relevance. Conclusions: The CNN demonstrated expert-level performance regardless of image quality. Future studies may investigate use of smartphone cameras and image sets with greater variance in image quality to further explore the influence of these parameters on model performance. Financial Disclosure(s): Proprietary or commercial disclosure may be found in the Footnotes and Disclosures at the end of this article.

The rapid progress of large language models (LLMs) driving generative artificial intelligence applications heralds the potential of opportunities in health care. We conducted a review up to April 2023 on Google Scholar, Embase, MEDLINE, and Scopus using the following terms: “large language models,” “generative artificial intelligence,” “ophthalmology,” “ChatGPT,” and “eye,” based on relevance to this review. From a clinical viewpoint specific to ophthalmologists, we explore from the different stakeholders’ perspectives—including patients, physicians, and policymakers—the potential LLM applications in education, research, and clinical domains specific to ophthalmology. We also highlight the foreseeable challenges of LLM implementation into clinical practice, including the concerns of accuracy, interpretability, perpetuating bias, and data security. As LLMs continue to mature, it is essential for stakeholders to jointly establish standards for best practices to safeguard patient safety. Financial Disclosure(s): Proprietary or commercial disclosure may be found in the Footnotes and Disclosures at the end of this article.

Purpose: To evaluate the performance of a deep learning (DL) algorithm for retinopathy of prematurity (ROP) screening in Nepal and Mongolia. Design: Retrospective analysis of prospectively collected clinical data. Participants: Clinical information and fundus images were obtained from infants in 2 ROP screening programs in Nepal and Mongolia. Methods: Fundus images were obtained using the Forus 3nethra neo (Forus Health) in Nepal and the RetCam Portable (Natus Medical, Inc.) in Mongolia. The overall severity of ROP was determined from the medical record using the International Classification of ROP (ICROP). The presence of plus disease was determined independently in each image using a reference standard diagnosis. The Imaging and Informatics for ROP (i-ROP) DL algorithm was trained on images from the RetCam to classify plus disease and to assign a vascular severity score (VSS) from 1 through 9. Main Outcome Measures: Area under the receiver operating characteristic curve and area under the precision-recall curve for the presence of plus disease or type 1 ROP and association between VSS and ICROP disease category. Results: The prevalence of type 1 ROP was found to be higher in Mongolia (14.0%) than in Nepal (2.2%; P

Summary: Succinate dehydrogenase (SDH)-deficient renal cell carcinoma represents a rare subtype of hereditary kidney cancer. Clinical diagnosis can be challenging and there is little evidence to guide systemic therapeutic options. We performed genomic profiling of a cohort of tumors through the analysis of whole genomes, transcriptomes, as well as flow cytometry and immunohistochemistry in order to gain a deeper understanding of their molecular biology. We find neutral evolution after early tumor activation with a lack of secondary driver events. We show that these tumors have epithelial derivation, possibly from the macula densa, a specialized paracrine cell of the renal juxtaglomerular apparatus. They subsequently develop into immune excluded tumors. We provide transcriptomic and protein expression evidence of a highly specific tumor marker, PAPPA2. These translational findings have implications for the diagnosis and treatment for this rare tumor subtype.

Purpose: Developing robust artificial intelligence (AI) models for medical image analysis requires large quantities of diverse, well-chosen data that can prove challenging to collect because of privacy concerns, disease rarity, or diagnostic label quality. Collecting image-based datasets for retinopathy of prematurity (ROP), a potentially blinding disease, suffers from these challenges. Progressively growing generative adversarial networks (PGANs) may help, because they can synthesize highly realistic images that may increase both the size and diversity of medical datasets. Design: Diagnostic validation study of convolutional neural networks (CNNs) for plus disease detection, a component of severe ROP, using synthetic data. Participants: Five thousand eight hundred forty-two retinal fundus images (RFIs) collected from 963 preterm infants. Methods: Retinal vessel maps (RVMs) were segmented from RFIs. PGANs were trained to synthesize RVMs with normal, pre-plus, or plus disease vasculature. Convolutional neural networks were trained, using real or synthetic RVMs, to detect plus disease from 2 real RVM test datasets. Main Outcome Measures: Features of real and synthetic RVMs were evaluated using uniform manifold approximation and projection (UMAP). Similarities were evaluated at the dataset and feature level using Fréchet inception distance and Euclidean distance, respectively. CNN performance was assessed via area under the receiver operating characteristic curve (AUC); AUCs were compared via bootstrapping and Delong’s test for correlated receiver operating characteristic curves. Confusion matrices were compared using McNemar’s chi-square test and Cohen’s κ value. Results: The CNN trained on synthetic RVMs showed a significantly higher AUC (0.971; P = 0.006 and P = 0.004) and classified plus disease more similarly to a set of 8 international experts (κ = 0.922) than the CNN trained on real RVMs (AUC = 0.934; κ = 0.701). Real and synthetic RVMs overlapped, by plus disease diagnosis, on the UMAP manifold, showing that synthetic images spanned the disease severity spectrum. Fréchet inception distance and Euclidean distances suggested that real and synthetic RVMs were more dissimilar to one another than real RVMs were to one another, further suggesting that synthetic RVMs were distinct from the training data with respect to privacy considerations. Conclusions: Synthetic datasets may be useful for training robust medical AI models. Furthermore, PGANs may be able to synthesize realistic data for use without protected health information concerns.

Purpose: Develop computer vision models for image-based differentiation of bacterial and fungal corneal ulcers and compare their performance against human experts. Design: Cross-sectional comparison of diagnostic performance. Participants: Patients with acute, culture-proven bacterial or fungal keratitis from 4 centers in South India. Methods: Five convolutional neural networks (CNNs) were trained using images from handheld cameras collected from patients with culture-proven corneal ulcers in South India recruited as part of clinical trials conducted between 2006 and 2015. Their performance was evaluated on 2 hold-out test sets (1 single center and 1 multicenter) from South India. Twelve local expert cornea specialists performed remote interpretation of the images in the multicenter test set to enable direct comparison against CNN performance. Main Outcome Measures: Area under the receiver operating characteristic curve (AUC) individually and for each group collectively (i.e., CNN ensemble and human ensemble). Results: The best-performing CNN architecture was MobileNet, which attained an AUC of 0.86 on the single-center test set (other CNNs range, 0.68–0.84) and 0.83 on the multicenter test set (other CNNs range, 0.75–0.83). Expert human AUCs on the multicenter test set ranged from 0.42 to 0.79. The CNN ensemble achieved a statistically significantly higher AUC (0.84) than the human ensemble (0.76; P < 0.01). CNNs showed relatively higher accuracy for fungal (81%) versus bacterial (75%) ulcers, whereas humans showed relatively higher accuracy for bacterial (88%) versus fungal (56%) ulcers. An ensemble of the best-performing CNN and best-performing human achieved the highest AUC of 0.87, although this was not statistically significantly higher than the best CNN (0.83; P = 0.17) or best human (0.79; P = 0.09). Conclusions: Computer vision models achieved superhuman performance in identifying the underlying infectious cause of corneal ulcers compared with cornea specialists. The best-performing model, MobileNet, attained an AUC of 0.83 to 0.86 without any additional clinical or historical information. These findings suggest the potential for future implementation of these models to enable earlier directed antimicrobial therapy in the management of infectious keratitis, which may improve visual outcomes. Additional studies are ongoing to incorporate clinical history and expert opinion into predictive models.

Purpose: To compare the efficacy and efficiency of training neural networks for medical image classification using comparison labels indicating relative disease severity versus diagnostic class labels from a retinopathy of prematurity (ROP) image dataset. Design: Evaluation of diagnostic test or technology. Participants: Deep learning neural networks trained on expert-labeled wide-angle retinal images obtained from patients undergoing diagnostic ROP examinations obtained as part of the Imaging and Informatics in ROP (i-ROP) cohort study. Methods: Neural networks were trained with either class or comparison labels indicating plus disease severity in ROP retinal fundus images from 2 datasets. After training and validation, all networks underwent evaluation using a separate test dataset in 1 of 2 binary classification tasks: normal versus abnormal or plus versus nonplus. Main Outcome Measures: Area under the receiver operating characteristic curve (AUC) values were measured to assess network performance. Results: Given the same number of labels, neural networks learned more efficiently by comparison, generating significantly higher AUCs in both classification tasks across both datasets. Similarly, given the same number of images, comparison learning developed networks with significantly higher AUCs across both classification tasks in 1 of 2 datasets. The difference in efficiency and accuracy between models trained on either label type decreased as the size of the training set increased. Conclusions: Comparison labels individually are more informative and more abundant per sample than class labels. These findings indicate a potential means of overcoming the common obstacle of data variability and scarcity when training neural networks for medical image classification tasks.

Purpose: To determine whether handheld widefield OCT can be used to document retinopathy of prematurity (ROP) stage while using scleral depression to improve peripheral views. Design: Prospective, observational study. Participants: Consecutive neonates admitted to the neonatal intensive care unit in a single academic medical center who also met criteria for ROP screening and whose parents or guardians consented for them to undergo research imaging. Methods: Scleral depression was combined with widefield OCT using an investigational 400-kHz, 55° field of view, handheld OCT during routine ROP screening from October 28, 2020, through March 3, 2021. Main Outcome Measures: Acquisition of en face and B-scan imaging of the peripheral retina to assess early vitreoretinal pathologic features objectively, including the demarcation between vascularized and anterior avascular retina, the presence of early ridge formation, and small neovascular tufts. Results: Various stages of ROP were detected using a rapid-acquisition OCT system. In 1 neonate, serial OCT imaging over a 5-week period demonstrated accumulation of neovascular tufts with progression to stage 3 ROP with extraretinal fibrovascular proliferation along the ridge. Videography of this technique is included in this report for instructional purposes. Conclusions: Serial examinations using widefield OCT and scleral depression are feasible and may improve detection and documentation of ROP disease progression. Earlier detection of ROP-related proliferation may prevent vitreoretinal traction, retinal detachment, and blindness.

Purpose: Generative adversarial networks (GANs) are deep learning (DL) models that can create and modify realistic-appearing synthetic images, or deepfakes, from real images. The purpose of our study was to evaluate the ability of experts to discern synthesized retinal fundus images from real fundus images and to review the current uses and limitations of GANs in ophthalmology. Design: Development and expert evaluation of a GAN and an informal review of the literature. Participants: A total of 4282 image pairs of fundus images and retinal vessel maps acquired from a multicenter ROP screening program. Methods: Pix2Pix HD, a high-resolution GAN, was first trained and validated on fundus and vessel map image pairs and subsequently used to generate 880 images from a held-out test set. Fifty synthetic images from this test set and 50 different real images were presented to 4 expert ROP ophthalmologists using a custom online system for evaluation of whether the images were real or synthetic. Literature was reviewed on PubMed and Google Scholars using combinations of the terms ophthalmology, GANs, generative adversarial networks, ophthalmology, images, deepfakes, and synthetic. Ancestor search was performed to broaden results. Main Outcome Measures: Expert ability to discern real versus synthetic images was evaluated using percent accuracy. Statistical significance was evaluated using a Fisher exact test, with P values ≤ 0.05 thresholded for significance. Results: The expert majority correctly identified 59% of images as being real or synthetic (P = 0.1). Experts 1 to 4 correctly identified 54%, 58%, 49%, and 61% of images (P = 0.505, 0.158, 1.000, and 0.043, respectively). These results suggest that the majority of experts could not discern between real and synthetic images. Additionally, we identified 20 implementations of GANs in the ophthalmology literature, with applications in a variety of imaging modalities and ophthalmic diseases. Conclusions: Generative adversarial networks can create synthetic fundus images that are indiscernible from real fundus images by expert ROP ophthalmologists. Synthetic images may improve dataset augmentation for DL, may be used in trainee education, and may have implications for patient privacy.

Purpose: Retinopathy of prematurity (ROP) is a leading cause of childhood blindness related to oxygen exposure in premature infants. Since oxygen monitoring protocols have reduced the incidence of treatment-requiring ROP (TR-ROP), it remains unclear whether oxygen exposure remains a relevant risk factor for incident TR-ROP and aggressive ROP (A-ROP), a severe, rapidly progressing form of ROP. The purpose of this proof-of-concept study was to use electronic health record (EHR) data to evaluate early oxygen exposure as a predictive variable for developing TR-ROP and A-ROP. Design: Retrospective cohort study. Participants: Two hundred forty-four infants screened for ROP at a single academic center. Methods: For each infant, oxygen saturations and fraction of inspired oxygen (FiO2) were extracted manually from the EHR until 31 weeks postmenstrual age (PMA). Cumulative minimum, maximum, and mean oxygen saturation and FiO2 were calculated on a weekly basis. Random forest models were trained with 5-fold cross-validation using gestational age (GA) and cumulative minimum FiO2 at 30 weeks PMA to identify infants who developed TR-ROP. Secondary receiver operating characteristic (ROC) curve analysis of infants with or without A-ROP was performed without cross-validation because of small numbers. Main Outcome Measures: For each model, cross-validation performance for incident TR-ROP was assessed using area under the ROC curve (AUC) and area under the precision-recall curve (AUPRC) scores. For A-ROP, we calculated AUC and evaluated sensitivity and specificity at a high-sensitivity operating point. Results: Of the 244 infants included, 33 developed TR-ROP, of which 5 developed A-ROP. For incident TR-ROP, random forest models trained on GA plus cumulative minimum FiO2 (AUC = 0.93 ± 0.06; AUPRC = 0.76 ± 0.08) were not significantly better than models trained on GA alone (AUC = 0.92 ± 0.06 [P = 0.59]; AUPRC = 0.74 ± 0.12 [P = 0.32]). Models using oxygen alone showed an AUC of 0.80 ± 0.09. ROC analysis for A-ROP found an AUC of 0.92 (95% confidence interval, 0.87–0.96). Conclusions: Oxygen exposure can be extracted from the EHR and quantified as a risk factor for incident TR-ROP and A-ROP. Extracting quantifiable clinical features from the EHR may be useful for building risk models for multiple diseases and evaluating the complex relationships among oxygen exposure, ROP, and other sequelae of prematurity.

Background: The Canadian Society for International Health (CSIH) has developed a systematic approach to support the introduction of high-quality evidence-based clinical practice guidelines (CPGs) in Kazakhstan. We describe a process enabling the large-scale selection and adaptation of high-quality international CPGs requisite to the project, and how these are applied at the points of care to achieve tangible results in terms of improved patient care. Methods: The CPG adaptation and implementation process used was both pragmatic and useful to national contexts and needs. The strategies used robust methods reported in the international medical literature; CSIH provided additional technical support, capacity-building, and customised instruments to aid implementation. The process involved clear assignment of tasks and roles between the partners, emphasising accountability throughout the adaptation and implementation process. Findings: During the 3-year partnership, 100 CPG candidates were identified, 42 of these were adapted for use in Kazakhstan, and implementation of three of these CPGs began. Key challenges at the time of adaptation included obtaining permission to use and translate guidelines and local partners' limited experience with the principles of evidence-based medicine. Barriers to implementation included contradiction between evidence-based CPGs and national clinical protocols, little capacity to perform required tests, and physicians' and patients' attitudes. Interpretation: A preliminary review showed much success with the CPG implementation process at each of the three pilot sites. The monitoring process was a new approach for the implementation sites. It was determined that this approach to CPG implementation could extrapolate to activities at other sites in Kazakhstan. Funding: None.

Purpose: The International Classification of Retinopathy of Prematurity is a consensus statement that creates a standard nomenclature for classification of retinopathy of prematurity (ROP). It was initially published in 1984, expanded in 1987, and revisited in 2005. This article presents a third revision, the International Classification of Retinopathy of Prematurity, Third Edition (ICROP3), which is now required because of challenges such as: (1) concerns about subjectivity in critical elements of disease classification; (2) innovations in ophthalmic imaging; (3) novel pharmacologic therapies (e.g., anti-vascular endothelial growth factor agents) with unique regression and reactivation features after treatment compared with ablative therapies; and (4) recognition that patterns of ROP in some regions of the world do not fit neatly into the current classification system., Design: Review of evidence-based literature, along with expert consensus opinion., Participants: International ROP expert committee assembled in March 2019 representing 17 countries and comprising 14 pediatric ophthalmologists and 20 retinal specialists, as well as 12 women and 22 men., Methods: The committee was initially divided into 3 subcommittees-acute phase, regression or reactivation, and imaging-each of which used iterative videoconferences and an online message board to identify key challenges and approaches. Subsequently, the entire committee used iterative videoconferences, 2 in-person multiday meetings, and an online message board to develop consensus on classification., Main Outcome Measures: Consensus statement., Results: The ICROP3 retains current definitions such as zone (location of disease), stage (appearance of disease at the avascular-vascular junction), and circumferential extent of disease. Major updates in the ICROP3 include refined classification metrics (e.g., posterior zone II, notch, subcategorization of stage 5, and recognition that a continuous spectrum of vascular abnormality exists from normal to plus disease). Updates also include the definition of aggressive ROP to replace aggressive-posterior ROP because of increasing recognition that aggressive disease may occur in larger preterm infants and beyond the posterior retina, particularly in regions of the world with limited resources. ROP regression and reactivation are described in detail, with additional description of long-term sequelae., Conclusions: These principles may improve the quality and standardization of ROP care worldwide and may provide a foundation to improve research and clinical care., (Published by Elsevier Inc.)