Transparency in Artificial Intelligence Reporting in Ophthalmology-A Scoping Review.

Topic: This scoping review summarizes artificial intelligence (AI) reporting in ophthalmology literature in respect to model development and validation. We characterize the state of transparency in reporting of studies prospectively validating models for disease classification. Clinical Relevance: Understanding what elements authors currently describe regarding their AI models may aid in the future standardization of reporting. This review highlights the need for transparency to facilitate the critical appraisal of models prior to clinical implementation, to minimize bias and inappropriate use. Transparent reporting can improve effective and equitable use in clinical settings. Methods: Eligible articles (as of January 2022) from PubMed, Embase, Web of Science, and CINAHL were independently screened by 2 reviewers. All observational and clinical trial studies evaluating the performance of an AI model for disease classification of ophthalmic conditions were included. Studies were evaluated for reporting of parameters derived from reporting guidelines (CONSORT-AI, MI-CLAIM) and our previously published editorial on model cards. The reporting of these factors, which included basic model and dataset details (source, demographics), and prospective validation outcomes, were summarized. Results: Thirty-seven prospective validation studies were included in the scoping review. Eleven additional associated training and/or retrospective validation studies were included if this information could not be determined from the primary articles. These 37 studies validated 27 unique AI models; multiple studies evaluated the same algorithms (EyeArt, IDx-DR, and Medios AI). Details of model development were variably reported; 18 of 27 models described training dataset annotation and 10 of 27 studies reported training data distribution. Demographic information of training data was rarely reported; 7 of the 27 unique models reported age and gender and only 2 reported race and/or ethnicity. At the level of prospective clinical validation, age and gender of populations was more consistently reported (29 and 28 of 37 studies, respectively), but only 9 studies reported race and/or ethnicity data. Scope of use was difficult to discern for the majority of models. Fifteen studies did not state or imply primary users. Conclusion: Our scoping review demonstrates variable reporting of information related to both model development and validation. The intention of our study was not to assess the quality of the factors we examined, but to characterize what information is, and is not, regularly reported. Our results suggest the need for greater transparency in the reporting of information necessary to determine the appropriateness and fairness of these tools prior to clinical use. Financial Disclosures: Proprietary or commercial disclosure may be found in the Footnotes and Disclosures at the end of this article.

A Review of Cost-Effectiveness Analyses for Open Angle Glaucoma Management.

PURPOSE: Cost-effectiveness analyses (CEAs) quantify and compare both costs and measures of efficacy for different interventions. As the costs of glaucoma management to patients, payers, and physicians are increasing, we seek to investigate the role of CEAs in the field of glaucoma and how such studies impact clinical management. METHODS: We adhered to the "Preferred Reporting Items for Systematic Reviews and Meta-analyses" guidelines for our systematic review structure. Eligible studies included any full-text articles that investigated cost-effectiveness or cost-utility as it relates to the field of open angle glaucoma management in the United States. Risk of bias assessment was conducted using the validated Joanna Briggs Institute Critical Appraisal Checklist for Economic Evaluations. RESULTS: Eighteen studies were included in the review. Dates of publication ranged from 1983 to 2021. Most of the studies were published in the 2000s and performed CEAs in the domains of treatment/therapy, screening, and adherence for patients with primary angle open glaucoma. Of the 18 articles included, 14 focused on treatment, 2 on screening, and 2 on adherence. Most of these studies focused on the cost-effectiveness of different topical medical therapies, whereas only a few studies explored laser procedures, surgical interventions, or minimally invasive procedures. Economic models using decision analysis incorporating state-transition Markov cycles or Montecarlo simulations were widely used, however, the methodology among studies was variable, with a wide spectrum of inputs, measures of outcomes, and time horizons used. CONCLUSION: Overall, we found that cost-effectiveness research in glaucoma in the United States remains relatively unstructured, resulting in unclear and conflicting implications for clinical management.

Federated AI, Current State, and Future Potential.

Artificial intelligence and machine learning applications are becoming increasingly popular in health care and medical devices. The development of accurate machine learning algorithms requires large quantities of good and diverse data. This poses a challenge in health care because of the sensitive nature of sharing patient data. Decentralized algorithms through federated learning avoid data aggregation. In this paper we give an overview of federated learning, current examples in healthcare and ophthalmology, challenges, and next steps.

Applications of Artificial Intelligence and Deep Learning in Glaucoma.

Diagnosis and detection of progression of glaucoma remains challenging. Artificial intelligence-based tools have the potential to improve and standardize the assessment of glaucoma but development of these algorithms is difficult given the multimodal and variable nature of the diagnosis. Currently, most algorithms are focused on a single imaging modality, specifically screening and diagnosis based on fundus photos or optical coherence tomography images. Use of anterior segment optical coherence tomography and goniophotographs is limited. The majority of algorithms designed for disease progression prediction are based on visual fields. No studies in our literature search assessed the use of artificial intelligence for treatment response prediction and no studies conducted prospective testing of their algorithms. Additional challenges to the development of artificial intelligence-based tools include scarcity of data and a lack of consensus in diagnostic criteria. Although research in the use of artificial intelligence for glaucoma is promising, additional work is needed to develop clinically usable tools.

Home Monitoring of Glaucoma Using a Home Tonometer and a Novel Virtual Reality Visual Field Device: Acceptability and Feasibility.

OBJECTIVE: Our aim was to assess the acceptability and feasibility of iCare HOME tonometer (HT) and Virtual Field (VF) devices in the home monitoring of glaucoma. DESIGN: Prospective feasibility and acceptability study. SUBJECTS: Twenty patients (39 eyes) with primary open-angle glaucoma, open-angle glaucoma, ocular hypertension, or suspected glaucoma. METHODS: Patients were trained and instructed to bring 2 devices home for 1 week and use the HT 4 times/day for 4 days and the VF 3 times total. MAIN OUTCOME MEASURES: For acceptability, we conducted satisfaction surveys and semistructured, qualitative interviews with a thematic analysis. Feasibility was assessed by device usage and quality of tests. RESULTS: Most patients (73.7%) felt that the HT was easy to use, and 100% of them found the HT useful. All patients (100%) felt that VF was easy to use, and 94.4% of them found the VF useful. All patients (100%) obtained acceptable intraocular pressure and completed a VF test at home. We identified 4 key themes, with 33 subthemes. The key themes include the following: (1) advantages of home monitoring; (2) difficulties with home monitoring; (3) future considerations in home monitoring; and (4) the experience of patients with glaucoma. CONCLUSIONS: The HT and VF were acceptable and feasible in a small cohort of motivated subjects. Patients were able to perform these tests proficiently at home, and they were generally enthused to obtain more data about their intraocular health, as it allowed them a heightened sense of security and insight about their chronic disease, as well as a reduction in foreseeable barriers to care. Home monitoring may also improve upon glaucoma care by enhancing patient empowerment and fostering community bonds. The VF should be further evaluated to ensure validity. FINANCIAL DISCLOSURE(S): Proprietary or commercial disclosure may be found after the references.

Intraocular Metastasis of Large T-cell Lymphoma Transformed from Mycosis Fungoides.

PURPOSE: To describe a rare case of intraocular lymphoma that metastasized from cutaneous mycosis fungoides and transformed to large cell T cell lymphoma resulting in vitreoretinal pathology. METHODS: Retrospective case report. RESULTS: A 57-year-old male presented with 3 months of blurred vision in the right eye. He reported only a medical history of psoriasis. Examination revealed keratic precipitates and dense vitritis in the right eye. He was taken for a diagnostic vitrectomy. Histopathology showed that atypical lymphoid cells and flow cytometry were consistent with transformed large cell T-cell lymphoma. During follow-up, pre- and inner retinal lesions were noted throughout the posterior pole. Histopathology of the psoriatic lesions was consistent with mycosis fungoides. He was initiated on systemic and intravitreal methotrexate with improvement in vision. CONCLUSIONS: Ocular involvement in metastatic transformed T-cell lymphoma is extremely rare but can be present with vitritis and retinal deposits. Our patient responded well to intravitreal methotrexate therapy.

Methods and Impact for Using Federated Learning to Collaborate on Clinical Research.

BACKGROUND: The development of accurate machine learning algorithms requires sufficient quantities of diverse data. This poses a challenge in health care because of the sensitive and siloed nature of biomedical information. Decentralized algorithms through federated learning (FL) avoid data aggregation by instead distributing algorithms to the data before centrally updating one global model. OBJECTIVE: To establish a multicenter collaboration and assess the feasibility of using FL to train machine learning models for intracranial hemorrhage (ICH) detection without sharing data between sites. METHODS: Five neurosurgery departments across the United States collaborated to establish a federated network and train a convolutional neural network to detect ICH on computed tomography scans. The global FL model was benchmarked against a standard, centrally trained model using a held-out data set and was compared against locally trained models using site data. RESULTS: A federated network of practicing neurosurgeon scientists was successfully initiated to train a model for predicting ICH. The FL model achieved an area under the ROC curve of 0.9487 (95% CI 0.9471-0.9503) when predicting all subtypes of ICH compared with a benchmark (non-FL) area under the ROC curve of 0.9753 (95% CI 0.9742-0.9764), although performance varied by subtype. The FL model consistently achieved top three performance when validated on any site's data, suggesting improved generalizability. A qualitative survey described the experience of participants in the federated network. CONCLUSION: This study demonstrates the feasibility of implementing a federated network for multi-institutional collaboration among clinicians and using FL to conduct machine learning research, thereby opening a new paradigm for neurosurgical collaboration.

Validation of the New York University Langone Eye Test Application, a Smartphone-Based Visual Acuity Test.

Purpose: To validate and assess user satisfaction and usability of the New York University (NYU) Langone Eye Test application, a smartphone-based visual acuity (VA) test. Design: Mixed-methods cross-sectional cohort study. Participants: Two hundred forty-four eyes of 125 participants were included. All participants were adults 18 years of age or older. Participants' eyes with a VA of 20/400 (1.3 logarithm of the minimum angle of resolution [logMAR]) or worse were excluded. Methods: Patients were tested using the clinical standard Rosenbaum near card and the NYU Langone Eye Test application on an iPhone and Android device. Each test was performed twice to measure reliability. Ten patients were selected randomly for subsequent semistructured qualitative interviews with thematic analysis. Main Outcome Measures: Visual acuity was the parameter measured. Bland-Altman analysis was used to measure agreement between the results of the NYU Langone Eye Test application and Rosenbaum card, as well as test-retest reliability of each VA. The correlation between results was calculated using the intraclass correlation coefficient. Satisfaction survey and semistructured interview questions were developed to measure usability and acceptability. Results: Bland-Altman analysis revealed an agreement between the application and the Rosenbaum near card of 0.017 ± 0.28 logMAR (iPhone) and 0.009 ± 0.29 logMAR (Android). The correlation between the application and the Rosenbaum near card was 0.74 for both the iPhone and Android. Test-retest reliability was 0.003 ± 0.22 logMAR (iPhone), 0.01 ± 0.25 logMAR (Android), and 0.01 ± 0.23 logMAR (Rosenbaum card). Of the 125 participants, 97.6% found the application easy to use, and 94.3% were overall satisfied with the application. Thematic analysis yielded 6 key themes: (1) weaknesses of application, (2) benefits of the application, (3) tips for application improvement, (4) difficulties faced while using the application, (5) ideal patient for application, and (6) comparing application with traditional VA testing. Conclusions: The NYU Langone Eye Test application is a user-friendly, accurate, and reliable measure of near VA. The application's integration with the electronic health record, accessibility, and easy interpretation of results, among other features, make it ideal for telemedicine use.

Qualitative and quantitative comparison of ERGs with contact lens and adhesive skin electrodes.

PURPOSE: Traditional ERGs recorded using corneal electrodes can be difficult for some patients to tolerate. In the last several years, adhesive skin electrodes have gained in acceptance. In this report we present a qualitative comparison of waveforms as well as a quantitative analysis of correlation of amplitudes and implicit times of simultaneous ERG recordings using contact lens and skin electrodes. METHODS: 89 subjects were included; all were referred for full-field ERG testing for multiple indications. ERGs (obtained according to ISCEV standards) were recorded simultaneously from both eyes with ERG-jet corneal contact lens electrodes and LKC Technologies Sensor Strip adhesive skin electrodes using multi-channel instrumentation (Diagnosys LLC, Espion3). Waveforms, a-wave and b-wave amplitudes and implicit times were compared. RESULTS: Waveform morphologies were similar between electrode types. Regression coefficients (conversion factors) for a-wave and b-wave amplitudes under both photopic and scotopic conditions were tightly clustered. Regression coefficients for implicit times were nearly equal to 1.0. The regression coefficient for the entire amplitude dataset was 0.349, with an overall correlation of 0. 869 between amplitude recorded with skin and contact lens electrodes. The regression coefficient for the entire implicit time dataset was 0.967, with an overall correlation of 0.964 between skin and contact lens electrodes. CONCLUSIONS: Our best estimate for the conversion factor between ERG amplitudes recorded with adhesive skin electrodes and contact lens electrodes is 0.349-amplitudes with skin electrodes are about 1/3 the amplitudes recorded simultaneously from the same eyes with contact lens electrodes, with a high correlation. Implicit times are nearly identical for the two electrode types.

Ophthalmic imaging for the diagnosis and monitoring of glaucoma: A review.

The last three decades have been a turning point for early glaucoma diagnosis and monitoring by newly developed imaging techniques. Improvements in imaging of the anterior segment (AS), retinal nerve fibre layer (RNFL), optic nerve head (ONH), and macular ganglion cell complex (GCC) came into existence and broke new ground for early diagnosis and progression survey. This review will summarise the different ophthalmic imaging techniques, their principles, pros and cons, ongoing development processes, and the future of imaging methods. The aim is to provide the clinician with an understanding of the role of imaging methods and provide insight for appropriate use and evaluation of their outputs in glaucoma diagnosis and treatment.

The Impact of International Pricing Index Models on Anti-Vascular Endothelial Growth Factor (VEGF) Drug Costs in the United States.

OBJECTIVE: To evaluate the impact of three international pricing index models on Medicare Part B spending for intravitreal anti-vascular endothelial growth factor (VEGF) drugs Design: Cost analysis Methods: U.S. and international sales data from the Multinational Integrated Data Analysis (MIDAS) database was used with data from the U.S. Centers for Medicare and Medicaid Services (CMS) to calculate Medicare Part B spending on anti-VEGF drugs Main Outcome: Medicare Part B expenditures of anti-VEGF drugs under various international pricing index models Results: Total Medicare Part B savings was greatest (75%) under the "most favored nation" proposal to peg the U.S. price to the lowest international price. Under the "most favored nation" proposal, prices of aflibercept are reduced from $1825.80 to $507.17, bevacizumab from $74.39 to $27.55, and ranibizumab (3 units or 0.3mg) from $1057.08 to $99.72. CONCLUSION: International pricing index models are one of many pricing strategies that could lead to savings in Medicare Part B costs.

Pseudopemphigoid: Sarcoidosis presenting as cicatricial conjunctivitis with symblepharon.

Conjunctival involvement in sarcoidosis is commonly characterized by epibulbar nodules or follicular conjunctivitis. The authors describe an apparently healthy woman who developed extensive monocular cicatricial conjunctivitis with symblepharon. The array of conditions presenting with cicatricial conjunctivitis was considered, with mucous membrane pemphigoid leading the diagnostic possibilities. Conjunctival biopsy disclosed the non-infectious, non-caseating granulomas of sarcoidosis and a systemic evaluation disclosed pulmonary nodules and hilar lymphadenopathy. As the patient had no respiratory symptoms and an old history of hepatic steatosis, oral hydroxychloroquine and topical cyclosporin were chosen for therapy rather than systemic corticosteroids.

Visual Search in Amblyopia: Abnormal Fixational Eye Movements and Suboptimal Sampling Strategies.

Purpose: Microsaccades shift the image on the fovea and counteract visual fading. They are also thought to serve as an optimal sampling strategy while viewing complex visual scenes. The goal of our study was to assess visual search in amblyopic children. Methods: Twenty-one amblyopic children with varying severity of amblyopia and 10 healthy controls were recruited. Eye movements were recorded using infrared video-oculography during amblyopic and fellow eye viewing while the subjects performed (1) visual fixation, (2) exploration of a blank scene, and (3) visual search task (spot the difference between two images). The number of correctly identified picture differences and reaction time were recorded. Microsaccade, saccades, and intersaccadic drifts were analyzed in patients without latent nystagmus (LN). Slow phase velocities were computed for patients with LN. Results: Both patients with and without LN were able to spot the same number of differences but took longer during fellow eye viewing compared to controls. The ability to identify differences was diminished during amblyopic eye viewing particularly those with LN and severe amblyopia. We found reduced frequencies of microsaccades and saccades in both amblyopic and fellow eyes during fixation and visual search but not during exploration of blank scene. Across all tasks, amblyopes with LN had increased intersaccadic drifts. Conclusions: Our findings suggest that deficient microsaccade and saccadic activity contributes to poorer sampling strategy in amblyopia, which is seen in both amblyopic and fellow eye. These deficits are more notable among subjects who experienced binocular decorrelation earlier in life, with subsequent development of LN.

Association of Postpartum Predischarge Depot-Medroxyprogesterone Acetate with In-Hospital Breastfeeding Initiation.

OBJECTIVE: The primary aim of this study was to examine the association between postpartum predischarge depot-medroxyprogesterone acetate (ppDMPA) and in-hospital breastfeeding initiation, and the secondary exploratory aim was to determine if any racial disparities are associated with ppDMPA receipt or its relationship to breastfeeding initiation. MATERIALS AND METHODS: We conducted a cross-sectional retrospective chart review of maternal and newborn records at a large urban academic medical center. Variables extracted included in-hospital feeding choice, obstetrical and sociodemographic variables, infant characteristics, and ppDMPA receipt. The association of ppDMPA and maternal-child characteristics with breastfeeding initiation was examined using logistic regression analysis. RESULTS: Among singleton live births of 919 mother-infant pairs (76.5% African American [AA]), 67% initiated breastfeeding (34% exclusive and 33% mixed) and 31.4% received ppDMPA. Breastfeeding rates differed significantly between AA (60.7%) and non-AA mothers (86.6%), and ppDMPA also differed significantly between AA (37.6%) and non-AA mothers (11.6%). Adjusting for other independent predictors, mothers who received ppDMPA were 1.5 times more likely not to initiate breastfeeding if AA, and 5.2 times more likely not to initiate breastfeeding if non-AA. CONCLUSIONS: ppDMPA receipt was independently associated with decreased rates of breastfeeding initiation. Although more AA mothers received ppDMPA than non-AA, the association of ppDMPA with breastfeeding noninitiation was stronger in non-AA than in AA mothers. Future research should examine this question prospectively to ascertain if there is a cause-effect relationship and should address both physiological effects and social perceptions.

Emergency department characteristics and capabilities in Bogotá, Colombia.

BACKGROUND: Emergency departments (EDs) are a critical, yet heterogeneous, part of international emergency care. The National ED Inventories (NEDI) survey has been used in multiple countries as a standardized method to benchmark ED characteristics. We sought to describe the characteristics, resources, capabilities, and capacity of EDs in the densely populated capital city of Bogotá, Colombia. METHODS: Bogotá EDs accessible to the general public 24/7 were surveyed using the 23-item NEDI survey used in several other countries ( www.emnet-nedi.org ). ED staff were asked about ED characteristics with reference to calendar year 2011. RESULTS: Seventy EDs participated (82 % response). Most EDs (87 %) were located in hospitals, and 83 % were independent hospital departments. The median annual ED visit volume was approximately 50,000 visits. Approximately 90 % (95 % confidence interval (CI) 80-96 %) had a contiguous layout, with medical and surgical care provided in one area. Almost all EDs saw both adults and children (91 %), while 6 % saw only adults and 3 % saw only children. Availability of technological and consultant resources in EDs was variable. Nearly every ED had cardiac monitoring (99 %, 95 % CI 92-100 %), but less than half had a dedicated CT scanner (39 %, 95 % CI 28-52 %). While most EDs were able to treat trauma 24/7 (81 %, 95 % CI 69-89 %), few could manage oncological (22 %, 95 % CI 13-34 %) or dental (3 %, 95 % CI 0-11 %) emergencies 24/7. The typical ED length-of-stay was between 1 and 6 h in 59 % of EDs (95 % CI, 46-70 %), while most others reported that patients remained for >6 h (39 %). Almost half of respondents (46 %, 95 % CI 34-59 %) reported their ED was over capacity. CONCLUSIONS: Bogotá EDs have high annual visit volumes and long length-of-stay, and half are over capacity. To meet the emergency care needs of people in Bogotá and other large cities, Colombia should consider improving urban ED capacity and training more emergency medicine specialists capable of efficiently staffing its large and crowded EDs.