Telemedicine Chapter 15: Telemedicine and Opthalmology

This chapter is part of Literature reviews carried out for the Heath Service Executive National Telehealth Steering Group April – July 2020

Systematic Reviews

Islam, Md Mohaimenul et al (2020) [Systematic Review and MetaAnalysis] Deep learning algorithms for detection of diabetic retinopathy in retinal fundus photographs: A systematic review and meta-analysis [1]

Background: Diabetic retinopathy (DR)is one of the leading causes of blindness globally. Earlier detection and timely treatment of DR are desirable to reduce the incidence and progression of vision loss. Currently, deep learning (DL) approaches have offered better performance in detecting DR from retinal fundus images. We, therefore, performed a systematic review with a meta-analysis of relevant studies to quantify the performance of DL algorithms for detecting DR. Methods: A systematic literature search on EMBASE, PubMed, Google Scholar, Scopus was performed between January 1, 2000, and March 31, 2019. The search strategy was based on the Preferred Reporting Items for Systematic Reviews and Meta-analyses (PRISMA) reporting guidelines, and DL-based study design was mandatory for articles inclusion. Two independent authors screened abstracts and titles against inclusion and exclusion criteria. Data were extracted by two authors independently using a standard form and the Quality Assessment of Diagnostic Accuracy Studies (QUADAS-2)tool was used for the risk of bias and applicability assessment. Results: Twenty-three studies were included in the systematic review; 20 studies met inclusion criteria for the metaanalysis. The pooled area under the receiving operating curve (AUROC) of DR was 0.97 (95%CI: 0.95-0.98), sensitivity was 0.83 (95%CI: 0.83-0.83), and specificity was 0.92 (95%CI: 0.92-0.92). The positive- and negative-likelihood ratio were 14.11 (95%CI: 9.91-20.07), and 0.10 (95%CI: 0.07-0.16), respectively. Moreover, the diagnostic odds ratio for DL models was 136.83 (95%CI: 79.03- 236.93). All the studies provided a DR-grading scale, a human grader eg trained caregivers, ophthalmologists as a reference standard. Conclusion: The findings of our study showed that DL algorithms had high sensitivity and specificity for detecting referable DR from retinal fundus photographs. Applying a DL-based automated tool of assessing DR from color fundus images could provide an alternative solution to reduce misdiagnosis and improve workflow. A DL-based automated tool offers substantial benefits to reduce screening costs, accessibility to healthcare and ameliorate earlier treatments.

Tan, Choon Han et al (2020) [Literature Review and Meta-Analysis] Use of Smartphones to Detect Diabetic Retinopathy: Scoping Review and MetaAnalysis of Diagnostic Test Accuracy Studies [2]

Background: Diabetic retinopathy (DR), a common complication of diabetes mellitus, is the leading cause of impaired vision in adults worldwide. Smartphone ophthalmoscopy involves using a smartphone camera for digital retinal imaging. Utilizing smartphones to detect DR is potentially more affordable, accessible, and easier to use than conventional methods. Objective: This study aimed to determine the diagnostic accuracy of various smartphone ophthalmoscopy approaches for detecting DR in diabetic patients. Methods: We performed an electronic search for literature published from January 2000 to November 2018. We included studies involving diabetic patients, which compared the diagnostic accuracy of smartphone ophthalmoscopy for detecting DR to an accurate or commonly employed reference standard, such as indirect ophthalmoscopy, slit-lamp biomicroscopy, and tabletop fundus photography. Two reviewers independently screened studies against the inclusion criteria, extracted data, and assessed the quality of included studies using the Quality Assessment of Diagnostic Accuracy Studies-2 tool, with disagreements resolved via consensus. Sensitivity and specificity were pooled using the random effects model. A summary receiver operating characteristic (SROC) curve was constructed. This review is reported in line with the Preferred Reporting Items for a Systematic Review and Meta-analysis of Diagnostic Test Accuracy Studies guidelines. Results: In all, nine studies involving 1430 participants were included. Most studies were of high quality, except one study with limited applicability because of its reference standard. The pooled sensitivity and specificity for detecting any DR was 87% (95% CI 74%-94%) and 94% (95% CI 81%-98%); mild nonproliferative DR (NPDR) was 39% (95% CI 10%-79%) and 95% (95% CI 91%-98%); moderate NPDR was 71% (95% CI 57%-81%) and 95% (95% CI 88%-98%); severe NPDR was 80% (95% CI 49%-94%) and 97% (95% CI 88%-99%); proliferative DR (PDR) was 92% (95% CI 79%-97%) and 99% (95% CI 96%-99%); diabetic macular edema was 79% (95% CI 63%-89%) and 93% (95% CI 82%-97%); and referral-warranted DR was 91% (95% CI 86%-94%) and 89% (95% CI 56%-98%). The area under SROC curve ranged from 0.879 to 0.979. The diagnostic odds ratio ranged from 11.3 to 1225. Conclusions: We found heterogeneous evidence showing that smartphone ophthalmoscopy performs well in detecting DR. The diagnostic accuracy for PDR was highest. Future studies should standardize reference criteria and classification criteria and evaluate other available forms of smartphone ophthalmoscopy in primary care settings.

Kawaguchi, Atsushi et al (2018)[Systematic Review and Meta-Analysis] Tele-Ophthalmology for Age-Related Macular Degeneration and Diabetic Retinopathy Screening: A Systematic Review and Meta-Analysis [3]

Background: To synthesize high-quality evidence to compare traditional in-person screening and tele-ophthalmology screening. Methods: Only randomized controlled trials (RCTs) were included in this systematic review and meta-analysis. The intervention of interest was any type of teleophthalmology, including screening of diseases using remote devices. Studies involved patients receiving care from any trained provider via teleophthalmology, compared with those receiving equivalent face-to-face care. A search was executed on the following databases: Medline, EMBASE, EBM Reviews, Global Health, EBSCO-CINAHL, SCOPUS, ProQuest Dissertations and Theses Global, OCLC Papers First, and Web of Science Core Collection. Six outcomes of care for age-related macular degeneration (AMD), diabetic retinopathy (DR), or glaucoma were measured and analyzed. Results: Two hundred thirty-seven records were assessed at the full-text level; six RCTs fulfilled inclusion criteria and were included in this review. Four studies involved participants with diabetes mellitus, and two studies examined choroidal neovascularization in AMD. Only data of detection of disease and participation in the screening program were used for the meta-analysis. Tele-ophthalmology had a 14% higher odds to detect disease than traditional examination; however, the result was not statistically significant (n = 2,012, odds ratio: 1.14, 95% confidence interval (CI): 0.52-2.53, p = 0.74). Meta-analysis results show that odds of having DR screening in the teleophthalmology group was 13.15 (95% CI: 8.01-21.61; p < 0.001) compared to the traditional screening program. Conclusions: The current evidence suggests that tele-ophthalmology for DR and age-related macular degeneration is as effective as in-person examination and potentially increases patient participation in screening.

Tan, Irene J et al (2017)[Systematic Review] Real-time teleophthalmology versus face-to-face consultation: A systematic review [4]

Advances in imaging capabilities and the evolution of real-time teleophthalmology have the potential to provide increased coverage to areas with limited ophthalmology services. However, there is limited research assessing the diagnostic accuracy of face-to-face teleophthalmology consultation. This systematic review aims to determine if real-time teleophthalmology provides comparable accuracy to face-to-face consultation for the diagnosis of common eye health conditions. A search of PubMed, Embase, Medline and Cochrane databases and manual citation review was conducted on 6 February and 7 April 2016. Included studies involved real-time telemedicine in the field of ophthalmology or optometry, and assessed diagnostic accuracy against gold-standard face-to-face consultation. The revised quality assessment of diagnostic accuracy studies (QUADAS-2) tool assessed risk of bias. Results Twelve studies were included, with participants ranging from four to 89 years old. A broad number of conditions were assessed and include corneal and retinal pathologies, strabismus, oculoplastics and post-operative review. Quality assessment identified a high or unclear risk of bias in patient selection (75%) due to an undisclosed recruitment processes. The index test showed high risk of bias in the included studies, due to the varied interpretation and conduct of realtime teleophthalmology methods. Reference standard risk was overall low (75%), as was the risk due to flow and timing (75%). Conclusion In terms of diagnostic accuracy, real-time teleophthalmology was considered superior to face-to-face consultation in one study and comparable in six studies. Store-and-forward image transmission coupled with real-time videoconferencing is a suitable alternative to overcome poor Internet transmission speeds.

Athikarisamy, Sam E et al (2015) [Systematic Review] Screening for
retinopathy of prematurity (ROP)using wide-angle digital retinal
photography by non-ophthalmologists: a systematic review [5]

Retinopathy of prematurity (ROP)is one of the leading and preventable causes of blindness. The investigation of choice for diagnosing ROP is binocular indirect ophthalmoscope (BIO) done by ophthalmologists. Since the number of ophthalmologists available to do BIO examination is limited, especially in developing countries, there is a need for an alternate, cheap, reliable and feasible test. Telemedicine imaging with Digital Retinal Photography (DRP)is one such alternate diagnostic test which can be performed easily by non-ophthalmologists, with adequate training. Our objective was to conduct a systematic review to evaluate the accuracy of DRP performed by trained personnel [non-ophthalmologists]in diagnosing clinically significant ROP. Medline, EMBASE, CINAHL and Cochrane databases were searched independently by two authors. Eligible studies were assessed using the Quality Assessment of Diagnostic Accuracy Studies (QUADAS)-2, an evidence-based tool for the assessment of quality in systematic reviews of diagnostic accuracy studies. Six were included in the review (three prospective; N=120, three retrospective; N=579). Studies had methodological limitations on QUADAS-2. Because of the heterogeneity of studies, data could not be pooled to derive single-effect size estimates for sensitivity and specificity. The included studies reported sensitivity of 45.5-100% with the majority being more than 90%; specificity 61.7-99.8% with the majority being more than 90%, positive predictive value 61.5-96.6% and negative predictive value of 76.9-100% for diagnosing clinically significant ROP. We conclude that diagnostic accuracy of DRP must be established in prospective studies with adequate sample size where DRP is compared against the simultaneously performed BIO examination.

Shi, Lili et al (2015) [Systematic Review and Meta-Analysis] Telemedicine for detecting diabetic retinopathy: a systematic review and meta-analysis [6]

Objective: To determine the diagnostic accuracy of telemedicine in various clinical levels of diabetic retinopathy (DR) and diabetic macular oedema (DME). Methods: PubMed, EMBASE and Cochrane databases were searched for telemedicine and DR. The methodological quality of included studies was evaluated using the Quality Assessment for Diagnostic Accuracy Studies (QUADAS-2). Measures of sensitivity, specificity and other variables were pooled using a random effects model. Summary receiver operating characteristic curves were used to estimate overall test performance. Metaregression and subgroup analyses were used to identify sources of heterogeneity. Publication bias was evaluated using Stata V.12.0. Results: Twenty articles involving 1960 participants were included. Pooled sensitivity of telemedicine exceeded 80% in detecting the absence of DR, low- or highrisk proliferative diabetic retinopathy (PDR), it exceeded 70% in detecting mild or moderate non-proliferative diabetic retinopathy (NPDR), DME and clinically significant macular oedema (CSME) and was 53% (95% CI 45% to 62%)in detecting severe NPDR. Pooled specificity of telemedicine exceeded 90%, except in the detection of mild NPDR which reached 89% (95% CI 88% to 91%). Diagnostic accuracy was higher with digital images obtained through mydriasis than through non-mydriasis, and was highest when a wide angle (100-200°) was used compared with a narrower angle (45-60°, 30° or 35°)in detecting the absence of DR and the presence of mild NPDR. No potential publication bias was detected. Conclusions: The diagnostic accuracy of telemedicine using digital imaging in DR is overall high. It can be used widely for DR screening. Telemedicine based on the digital imaging technique that combines mydriasis with a wide angle field (100-200°)is the best choice in detecting the absence of DR and the presence of mild NPDR.

Abràmoff, Michael D et al (2020)[Literature Review] Automated and Computer-Assisted Detection, Classification, and Diagnosis of Diabetic Retinopathy [7]

Background: The introduction of artificial intelligence (AI)in medicine has raised significant ethical, economic, and scientific controversies. Introduction:Because an explicit goal of AI is to perform processes previously reserved for human clinicians and other health care personnel, there is justified concern about the impact on patient safety, efficacy, equity, and liability. Discussion: Systems for computer-assisted and fully automated detection, triage, and diagnosis of diabetic retinopathy (DR)from retinal images show great variation in design, level of autonomy, and intended use. Moreover, the degree to which these systems have been evaluated and validated is heterogeneous. We use the term DR AI system as a general term for any system that interprets retinal images with at least some degree of autonomy froma human grader. We put forth these standardized descriptors to form a means to categorize systems for computer-assisted and fully automated detection, triage, and diagnosis of DR. The components of the categorization system include level of device autonomy, intended use, level of evidence for diagnostic accuracy, and system design. Conclusion: There is currently minimal empirical basis to assert that certain combinations of autonomy, accuracy, or intended use are better or more appropriate than any other. Therefore, at the current stage of development of this document, we have been descriptive rather than prescriptive, and we treat the different categorizations as independent and organized along multiple axes.

Bilong, Yannick et al (2020)[Pilot Study] Smartphone-Assisted Glaucoma Screening in Patients With Type 2 Diabetes: a Pilot Study [8]

We aimed to determine true and false positives of glaucoma screening, relying solely on photos of the retina, taken with a smartphone. We performed a descriptive and analytical study on patients with type 2 diabetes at the National Obesity Centre, Yaoundé, Cameroon. Participating patients had retinal photography sessions using an iPhone 5s (iOS 10.3.3; Apple, Cupertino, CA) coupled to the Make in India Retinal Camera (MIIRetCam; MIIRetCam Inc., Coimbatore, TN, India). Obtained pictures of the retina were stored and transferred via the Internet to an ophthalmologist to assess glaucoma. Selected patients were then invited to undergo a conventional ophthalmological examination to confirm the diagnosis. A total of 395 patients were screened, 39 (including 20 women) were diagnosed with suspicion of glaucoma based on retinal photos, a prevalence rate of 9.87%. The following signs were found; Cup/Disc ratio (C/D) ≥0.5 in 64.1% (25/39), asymmetric C/D >0.2 in 35.9% (14/39), papillary haemorrhage in 10.2% (4/39) and retinal nerve fibre deficiency in 2.5% (1/39). Only 14 of 39 patients with suspicion of glaucoma were examined, giving a lost-to-followup rate of 64.1%. Chronic open-angle glaucoma was confirmed in 8 patients [true positives] and absent in 6 patients [false positives]. The prevalence of smartphone-detected glaucoma and lost-to-follow-up rates were high. So we need to improve this type of screening, with additional tests such as transpalpebral applanation tonometer and the smartphone Frequency Doubling Technique visual field combined with better education of patients to increase their adherence to follow-up.

Brady, Christopher J et al (2020) [Literature Review] Telemedicine for Age-Related Macular Degeneration [9]

Background:As the leading cause of vision loss in the United States, agerelated macular degeneration (AMD) would seem to be amenable to interventions that increase access to screening and management services for patients. AMD poses several unique challenges for telemedicine, however. The disease lacks clinical consensus on the effectiveness and costeffectiveness of screening the general population, and more complex imaging modalities may be required than for what has traditionally been used for diabetic retinopathy telehealth systems. Methods: The current literature was reviewed to find clinical trials and expert consensus documents on the state-of-the-art of telemedicine for AMD. Results:A range of feasibility studies have reported success with telemedicine strategies for AMD. Several investigators have reported experience with AMD screening and remote-monitoring systems as well as artificial intelligence applications. Conclusions: There are currently no large-scale telemedicine programs for either screening or managing AMD, but new approaches to screening and managing the condition may allow for expansion of highquality convenient care for an increasing patient population.

Brady, Christopher J et al (2020)[Literature Review] Telemedicine for Retinopathy of Prematurity [10]

Background: Retinopathy of prematurity (ROP)is a disease of the retinal vasculature that remains a leading cause of childhood blindness worldwide despite improvements in the systemic care of premature newborns. Screening for ROP is effective and cost-effective, but in many areas, access to skilled examiners to conduct dilated examinations is poor. Remote screening with retinal photography is an alternative strategy that may allow for improved ROP care. Methods: The current literature was reviewed to find clinical trials and expert consensus documents on the state-of-the-art of telemedicine for ROP. Results: Several studies have confirmed the utility of telemedicine for ROP. In addition, several clinical studies have reported favorable long-term results. Many investigators have reinforced the need for detailed protocols on image acquisition and image interpretation. Conclusions: Telemedicine for ROP appears to be a viable alternative to live ophthalmoscopic examinations in many circumstances. Standardization and documentation afforded by telemedicine may provide additional benefits to providers and their patients. With continued improvements in image quality and affordability of imaging systems as well as improved automated image interpretation tools anticipated in the near future, telemedicine for ROP is expected to play an expanding role for a uniquely vulnerable patient population.

Cai, Sophie et al (2020) [Literature Review] Recent developments in pediatric retina [11]

Purpose: Pediatric retina is an exciting, but also challenging field, where patient age and cooperation can limit ease of diagnosis of a broad range of congenital and acquired diseases, inherited retinal degenerations are mostly untreatable and surgical outcomes can be quite different from those for adults. This review aims to highlight some recent advances and trends that are improving our ability to care for children with retinal conditions. Recent Findings: Studies have demonstrated the feasibility of multimodal imaging even in nonsedated infants, with portable optical coherence tomography (OCT) and OCT angiography in particular offering structural insights into diverse pediatric retinal conditions. Encouraging long-term outcomes of subretinal voretigene neparvovec-rzyl injection for RPE65 mutationassociated Leber congenital amaurosis have inspired research on the optimization of subretinal gene delivery and gene therapy for other inherited retinal degenerations. In retinopathy of prematurity, machine learning and smartphone-based imaging can facilitate screening, and studies have highlighted favorable outcomes from intravitreal anti-vascular endothelial growth factor [anti-VEGF] injections. A nomogram for pediatric pars plana sclerotomy site placement may improve safety in complex surgeries. Summary:Multimodal imaging, gene therapy, machine learning and surgical innovation have been and will continue to be important to advances in pediatric retina.

Cao, Yingpin et al (2020)[Pilot Study] An effectiveness study of a wearable device (Clouclip)intervention in unhealthy visual behaviors among school-age children: A pilot study [12]

Introduction: The study aimed to determine the effectiveness of an intervention for unhealthy visual behaviors of school-age children using a wearable device (Clouclip). Method: The design was a self-controlled prospective study. Clouclip, with the vibration alert disabled, was first applied to measure baseline near-work behaviors in the first week. The vibration alert was then enabled to signal unhealthy visual behaviors (near-work distance < 30 cm and >5 seconds, or near-work distance <60 cm for >45 minutes)for 3 weeks. Near-work behaviors were measured again at the first week and the first month after intervention, respectively. The changes in behaviors between the baseline and the first week and the first month after intervention were analyzed. Results: Sixty-seven subjects were eligible for this experiment (the mean age 10.45 ± 0.50 years, 34 boys). Children who logged sufficient wearing time (12.30 ± 0.18 hours on weekdays and 12.16 ± 0.23 hours on weekends) were included for analysis. The average daily near-work distance was significantly increased after the vibration intervention. The time ratio of near-work activity <30 cm to the total <60 cm and the frequency of continuous near-work (distance <60 cm and continuous time >30 minutes) were significantly decreased after the intervention. Although some of the effects were reversed with time following the intervention, some were observed to be maintained until the end of the observation period, and the improvement of the behaviors was more prominent in children who had a shorter near-work distance (<30 cm) at baseline. Conclusions: In conclusion, Clouclip can significantly modify near-work behaviors in school-age children and it can last a certain period of time. If these behaviors are causes of myopia development and progression, Clouclip might provide a strategy for managing myopia.

Chun, Jaehyeong et al (2020)[Validation Study] Deep Learning-Based Prediction of Refractive Error Using Photorefraction Images Captured by a Smartphone: Model Development and Validation Study [13]

Background: Accurately predicting refractive error in children is crucial for detecting amblyopia, which can lead to permanent visual impairment, but is potentially curable if detected early. Various tools have been adopted to more easily screen a large number of patients for amblyopia risk. Objective: For efficient screening, easy access to screening tools and an accurate prediction algorithm are the most important factors. In this study, we developed an automated deep learning-based system to predict the range of refractive error in children (mean age 4.32 years, SD 1.87 years) using 305 eccentric photorefraction images captured with a smartphone. Methods: Photorefraction images were divided into seven classes according to their spherical values as measured by cycloplegic refraction. Results: The trained deep learning model had an overall accuracy of 81.6%, with the following accuracies for each refractive error class: 80.0% for ≤-5.0 diopters (D), 77.8% for >-5.0 D and ≤-3.0 D, 82.0% for >-3.0 D and ≤-0.5 D, 83.3% for >-0.5 D and <+0.5 D, 82.8% for ≥+0.5 D and <+3.0 D, 79.3% for ≥+3.0 D and <+5.0 D, and 75.0% for ≥+5.0 D. These results indicate that our deep learning-based system performed sufficiently accurately. Conclusions: This study demonstrated the potential of precise smartphone-based prediction systems for refractive error using deep learning and further yielded a robust collection of pediatric photorefraction images.

Fatehi, F et al (2020)[Literature Review] Teleophthalmology for the elderly population: A review of the literature [14]

Background: Ophthalmology is one of the most requested medical speciality services in the elderly population. Although numerous studies have shown the potentials of telemedicine for the provision of ophthalmology services, the extent of its usability in older adults and the aged population is not clear. The aim of this study was to investigate the characteristics and usability features of teleophthalmology for the elderly population. Method: We searched PubMed, Embase, Scopus and CINAHL for relevant studies since 2008. Forty-five papers met the eligibility criteria and included in this review. We used a multifaceted model to extract the data and analyze findings by cross-tabulation. Results: The majority of the reviewed papers included participants of 65 years of age or older. Most of the studies were conducted in the USA (38 %). Diabetic retinopathy, glaucoma, age-related macular degeneration and cataract were the most researched eye diseases, and among the imaging technologies, retinal photography had been used the most (72 %). The studies showed teleophthalmology can improve access to specialty care, reduce the number of unnecessary visits, alleviate overloads on treatment centers, and provide more comprehensive exams. It also made services cost-saving for stakeholders and cost-effective in rural areas. However, teleophthalmology was not cost-effective for patients above 80 and low-density population areas. Conclusion: Evidence is lacking for the usability and effectiveness of teleophthalmology for the elderly population. The findings suggest that primary care providers in collaboration with ophthalmologists could provide more effective eye care to elderly population. Appropriate training is also necessary for primary care doctors to manage and refer older patients in a timely manner. Diagnostic value and cost-effective imaging modalities which are the core of the teleophthalmology, can be enhanced by image processing techniques and artificial intelligence.

Gan, Kenman et al (2020)[Recommendations] Telemedicine for
Glaucoma: Guidelines and Recommendations [15]

Background: Glaucoma is the leading cause of irreversible blindness worldwide. Access to glaucoma specialists is challenging and likely to become more difficult as the population ages. Introduction:Using telemedicine for glaucoma has the potential to increase access to glaucoma care by improving efficiency and decreasing the need for long-distance travel for patients. Results: Teleglaucoma programs can be used for screening, diagnostic consultation, and long-term treatment monitoring. Key components of teleglaucoma programs include patient history, equipment, intraocular pressure measurement, pachymetry, anterior chamber imaging/gonioscopy, fundus photography, retinal nerve fiber layer imaging, medical record and imaging software, and skilled personnel. Discussion: Teleglaucoma has tremendous potential to improve patient access to highquality cost-effective glaucoma care. Conclusions: We have reviewed some special considerations needed to address the complexity of providing guideline-concordant glaucoma care.

Greenwald, Miles F et al (2020)[Evaluation Study] Evaluation of artificialnintelligence-based telemedicine screening for retinopathy of prematurity [16]

Retrospective evaluation of a deep learning-derived retinopathy of prematurity (ROP) vascular severity score in an operational ROP screening program demonstrated high diagnostic performance for detection of type 2 or worse ROP. To our knowledge, this is the first report in the literature that evaluated the use of artificial intelligence for ROP screening and represents a proof of concept. With further prospective validation, this technology might improve the accuracy, efficiency, and objectivity of diagnosis and facilitate earlier detection of disease progression in patients with potentially blinding ROP.

Hogarty, Daniel T et al (2020)[Literature Review] Smartphone use in ophthalmology: What is their place in clinical practice? [17]

Smartphones are an increasingly common and rapidly developing tool in clinical practice. Numerous applications or “apps” are available for use on smartphones that aim to help clinicians perform a variety of tasks at the point of care. A large number of ophthalmology-related medical apps that can perform a variety of clinically relevant functions are now available in virtual stores such as the Google Play™ Store or the Apple App Store®. On the ophthalmic front, these include measures of visual acuity, tools to assist in the assessment and treatment of conditions such as amblyopia and glaucoma, as well as add-on devices that allow visualization and photography of the anterior and posterior segments of the eye. Despite the large number of available programs, the evidence supporting their use is unclear, with issues concerning professional input in development, regulation, validation, and security of information. We present the various uses of smartphones in ophthalmology and summarize the current literature.

Horton, Mark B et al (2020)[Practice Guidelines] Practice Guidelines for Ocular Telehealth-Diabetic Retinopathy, Third Edition [18]

Contributors The following document and appendices represent the third edition of the Practice Guidelines for Ocular Telehealth-Diabetic Retinopathy. These guidelines were developed by the Diabetic Retinopathy Telehealth Practice Guidelines Working Group. This working group consisted of a large number of subject matter experts in clinical applications for telehealth in ophthalmology.

Kern, Christoph et al (2020)[Feasibility Study] Implementation of a cloudbased referral platform in ophthalmology: making telemedicine services a reality in eye care [19]

Background: Hospital Eye Services (HES)in the UK face an increasing number of optometric referrals driven by progress in retinal imaging. The National Health Service (NHS) published a 10-year strategy (NHS Long-Term Plan)to transform services to meet this challenge. In this study, we implemented a cloud-based referral platform to improve communication between optometrists and ophthalmologists. Methods: Retrospective cohort study conducted at Moorfields Eye Hospital, Croydon (NHS Foundation Trust, London, UK). Patients classified into the HES referral pathway by contributing optometrists have been included into this study. Main outcome measures was the reduction of unnecessary referrals. Results: After reviewing the patient’s data in a web-based interface 54 (52%) out of 103 attending patients initially classified into the referral pathway did not need a specialist referral. Fourteen (14%) patients needing urgent treatment were identified. Usability was measured in duration for data input and reviewing which was an average of 9.2 min (median: 5.4; IQR: 3.4-8.7)for optometrists and 3.0 min (median: 3.0; IQR: 1.7-3.9) min for ophthalmologists. A variety of diagnosis was covered by this tool with dry age-related macular degeneration (n=34) being most common. Conclusion: After implementation more than half of the HES referrals have been avoided. This platform offers a digital-first solution that enables rapid-access eye care for patients in community optometrists, facilitates communication between healthcare providers and may serve as a foundation for implementation of artificial intelligence.

Keskinbora, Kadircan et al (2020)[Literature Review] Artificial
Intelligence and Ophthalmology [20]

Artificial intelligence is advancing rapidly and making its way into all areas of our lives. This review discusses developments and potential practices regarding the use of artificial intelligence in the field of ophthalmology, and the related topic of medical ethics. Various artificial intelligence applications related to the diagnosis of eye diseases were researched in books, journals, search engines, print and social media. Resources were cross-checked to verify the information. Artificial intelligence algorithms, some of which were approved by the US Food and Drug Administration, have been adopted in the field of ophthalmology, especially in diagnostic studies. Studies are being conducted that prove that artificial intelligence algorithms can be used in the field of ophthalmology, especially in diabetic retinopathy, age-related macular degeneration, and retinopathy of prematurity. Some of these algorithms have come to the approval stage. The current point in artificial intelligence studies shows that this technology has advanced considerably and shows promise for future work. It is believed that artificial intelligence applications will be effective in identifying patients with preventable vision loss and directing them to physicians, especially in developing countries where there are fewer trained professionals and physicians are difficult to reach. When we consider the possibility that some future artificial intelligence systems may be candidates for moral/ethical status, certain ethical issues arise. Questions about moral/ethical status are important in some areas of applied ethics. Although it is accepted that current intelligence systems do not have moral/ethical status, it has yet to be determined what the exact the characteristics that confer moral/ethical status are or will be.

Kim, Tyson N et al (2020)[Feasibility Study] Comparison of automated and expert human grading of diabetic retinopathy using smartphonebased retinal photography [21]

Purpose: The aim of this study is to investigate the efficacy of a mobile platform that combines smartphone-based retinal imaging with automated grading for determining the presence of referral-warranted diabetic retinopathy (RWDR). Methods: A smartphone-based camera (RetinaScope) was used by non-ophthalmic personnel to image the retina of patients with diabetes. Images were analyzed with the Eyenuk EyeArt® system, which generated referral recommendations based on presence of diabetic retinopathy (DR) and/or markers for clinically significant macular oedema. Images were independently evaluated by two masked readers and categorized as refer/no refer. The accuracies of the graders and automated interpretation were determined by comparing results to gold standard clinical diagnoses. Results: A total of 119 eyes from 69 patients were included. RWDR was present in 88 eyes (73.9%) and in 54 patients (78.3%). At the patient-level, automated interpretation had a sensitivity of 87.0% and specificity of 78.6%; grader 1 had a sensitivity of 96.3% and specificity of 42.9%; grader 2 had a sensitivity of 92.5% and specificity of 50.0%. At the eye-level, automated interpretation had a sensitivity of 77.8% and specificity of 71.5%; grader 1 had a sensitivity of 94.0% and specificity of 52.2%; grader 2 had a sensitivity of 89.5% and specificity of 66.9%. Discussion: Retinal photography with RetinaScope combined with automated interpretation by EyeArt achieved a lower sensitivity but higher specificity than trained expert graders. Feasibility testing was performed using non-ophthalmic personnel in a retina clinic with high disease burden. Additional studies are needed to assess efficacy of screening diabetic patients from general population.

Koh, Joel En Wie et al (2020)[Pilot Study] A novel hybrid approach for automated detection of retinal detachment using ultrasound images [22]

Retinal detachment (RD)is an ocular emergency, which needs quick intervention to preclude permanent vision loss. In general, ocular ultrasound is used by ophthalmologists to enhance their judgment in detecting RD in eyes with media opacities which precludes the retinal evaluation. However, the quality of ultrasound (US)images may be degraded due to the presence of noise, and other retinal conditions may cause membranous echoes. All these can influence the accuracy of diagnosis. Hence, to overcome the above, we are proposing an automated system to detect RD using texton, higher order spectral (HOS) cumulants and locality sensitive discriminant analysis (LSDA) techniques. Our developed method is able to classify the posterior vitreous detachment and RD using support vector machine classifier with highest accuracy of 99.13%. Our system is ready to be tested with more diverse ultrasound images and aid ophthalmologists to arrive at a more accurate diagnosis.

Lam, Alexander et al(2020)Use of Virtual Reality Simulation to Identify Vision-Related Disability in Patients With Glaucoma [23]

Importance: Clinical assessment of vision-related disability is hampered by the lack of instruments to assess visual performance in real-world situations. Interactive virtual reality (VR) environments displayed in a binocular stereoscopic VR headset have been designed, presumably simulating day-to-day activities to evaluate vision-related disability. Objective: To investigate the application of VR to identify vision-related disability in patients with glaucoma. Design, Setting, and Participants: In a cross-sectional study, 98 patients with glaucoma and 50 healthy individuals were consecutively recruited from a university eye clinic; all participants were Chinese. The study was conducted between August 30, 2016, and July 31, 2017; data analysis was performed from December 1, 2017, to October 30, 208 18. Exposures: Measurements of visual acuity, contrast sensitivity, visual field (VF), National Eye Institute 25-item Visual Function Questionnaire Rasch score, and VR disability scores determined from 5 VR simulations: supermarket shopping, stair and city navigations in daytime, and stair and city navigations in nighttime. Duration required to complete the simulation, number of items incorrectly identified, and number of collisions were measured to compute task-specific and overall VR disability scores. Visionrelated disability was identified when the VR disability score was outside the normal age-adjusted 95% confidence region. Main Outcomes and Measures: Virtual reality disability score. Results: In the 98 patients with glaucoma, mean (SD) age was 49.8 (11.6) years and 60 were men (61.2%); in the 50 healthy individuals, mean (SD) age was 48.3 (14.8) years and 16 were men (32.0%). The patients with glaucoma had different degrees of VF loss (122 eyes [62.2%]had moderate or advanced VF defects). The time required to complete the activities by patients with glaucoma vs healthy individuals was longer by 15.2 seconds (95% CI, 5.5-24.9 seconds) or 34.1% (95% CI, 12.4%- 55.7%)for the shopping simulation, 72.8 seconds (95% CI, 23.0-122.6 seconds) or 33.8% (95% CI, 10.7%-56.9%)for the nighttime stair navigation, and 38.1 seconds (95% CI, 10.9-65.2 seconds) or 30.8% (95% CI, 8.8%-52.8%) for the nighttime city navigation. The mean (SD) duration was not significantly different between the glaucoma and healthy groups in daytime stair (203.7 [93.7] vs 192.9 [89.1] seconds, P = .52) and city (118.7 [41.5] vs 117.0 [52.3] seconds, P = .85) navigation. For each decibel decrease in binocular VF sensitivity, the risk of collision increased by 15% in nighttime stair (hazard ratio [HR], 1.15; 95% CI, 1.08-1.22) and city (HR, 1.15; 95% CI, 1.08-1.23) navigations. Fifty-eight patients (59.1%) with glaucoma had vision-related disability in at least 1 simulated daily task; a higher proportion of patients had vision-related disability in nighttime city (27 of 88 [30.7%]) and stair (27 of 90 [30.0%]) navigation than in daytime city (7 of 88 [8.0%]) and stair (19 of 96 [19.8%]) navigation. The overall VR disability score was associated with the National Eye Institute 25-item Visual Function Questionnaire Rasch score (R2 = 0.207). Conclusions and Relevance: These findings suggest that visionrelated disability is associated with lighting condition and task in patients with glaucoma. Virtual reality may allow eye care professionals to understand the patients’ perspectives on how visual impairment imparts disability in daily living and provide a new paradigm to augment the assessment of vision-related disability.

Lanzetta, Paolo et al (2020)[Recommendations] Fundamental principles of an effective diabetic retinopathy screening program [24]

Background: Diabetic retinopathy (DR)is the leading cause of blindness among working-age adults worldwide. Early detection and treatment are necessary to forestall vision loss from DR. Methods: A working group of ophthalmic and diabetes experts was established to develop a consensus on the key principles of an effective DR screening program. Recommendations are based on analysis of a structured literature review. Results: The recommendations for implementing an effective DR screening program are: 1. examination methods must be suitable for the screening region, and DR classification/grading systems must be systematic and uniformly applied. Two-field retinal imaging is sufficient for DR screening and is preferable to seven-field imaging, and referable DR should be well defined and reliably identifiable by qualified screening staff; 2. in many countries/regions, screening can and should take place outside the ophthalmology clinic; 3. screening staff should be accredited and show evidence of ongoing training; screening programs should adhere to relevant national quality assurance standards; 5. studies that use uniform definitions of risk to determine optimum risk-based screening intervals are required; 6. technology infrastructure should be in place to ensure that high-quality images can be stored securely to protect patient information; 7. although screening for diabetic macular edema (DME)in conjunction with DR evaluations may have merit, there is currently insufficient evidence to support implementation of programs solely for DME screening. Conclusion: Use of these recommendations may yield more effective DR screening programs that reduce the risk of vision loss worldwide.

Maa, April Y et al (2020)[Evaluation Study] Diagnostic Accuracy of Technology-based Eye Care Services: The Technology-based Eye Care Services Compare Trial Part I [25]

Purpose: Ophthalmologic telemedicine has the ability to provide eye care for patients remotely, and many countries have used screening tele-ophthalmology programs for several years. One such initiative at the Veterans Affairs (VA)Healthcare System is Technology-based Eye Care Services (TECS). The TECS services are located in primary care clinics and provide basic screening eye care, including vision, refraction, and retinal photography. Eye care providers (‘readers’) review the clinical data and recommend appropriate follow-up. One of the most common referrals from TECS has been for glaucoma, and this study was powered for glaucoma/glaucoma suspect detection. The current study was undertaken to identify aspects of the protocol that could be refined to enhance accuracy. Design: Prospective comparison between the standard TECS protocol versus a face-to-face (FTF) examination on 256 patients, all of whom had no known history of significant ocular disease. Participants: Patients with no known ocular disease who were scheduled for an in-person eye appointment at the Atlanta VA. Patients underwent screening through the TECS protocol and received an FTF examination on the same day [gold standard]. The TECS readers were masked to the results of the FTF examination. Main Outcome Measures: Percent agreement, kappa, sensitivity, and specificity were calculated for the TECS readers’ interpretations versus the FTF examination. Results: The TECS readers showed substantial agreement for cataract (κ ≥ 0.71) and diabetic retinopathy (κ≥ 0.61) and moderate to substantial agreement for glaucoma/glaucoma suspect (κ ≥ 0.52) compared with an FTF examination. Age-related macular degeneration (AMD) showed moderate agreement (κ ≥ 0.34). Percent agreement with the TECS protocol was high (84.3%-98.4%)for each of the disease categories. Overall sensitivity and specificity were ≥75% and ≥55%, respectively, for any diagnosis resulting in referral. Inter-reader and intra-reader agreement was substantial for most diagnoses (κ> 0.61) with percent agreements ranging from 66% to 99%. Conclusions: Our results indicate that the standard TECS protocol is accurate when compared with an FTF examination for the detection of common eye diseases. The inclusion of additional testing such as OCT could further enhance diagnostic capability.

Maa, April Y et al (2020)[Evaluation Study] The Impact of OCT on Diagnostic Accuracy of the Technology-Based Eye Care Services Protocol: Part II of the Technology-Based Eye Care Services Compare Trial [26]

Purpose: Ophthalmologic telemedicine programs help to address the growing demand for eye care and lessen healthcare disparities for patients. One example is Technology-Based Eye Care Services (TECS), implemented in the Veteran Affairs Healthcare System in 2015. Accuracy and quality data for TECS both have been reported, and data suggest that although the TECS examination is comparable with an in-person examination, sensitivity for glaucoma and glaucoma suspect detection is less than that for other diseases, such as macular degeneration. Several articles suggest that OCT can improve disease detection for glaucoma. Therefore, this study was undertaken to test the impact of OCT on the accuracy of the TECS protocol. This article reports the data from part II of the TECS Compare trial; results from part I are discussed in a previous article. Design: Prospective comparison between the TECS protocol with OCT versus a face-to-face (FTF) examination for 256 patients. Participants: An eligible patient was defined as a patient with no known ocular disease who desired a routine eye examination. Methods: Patient underwent the TECS protocol workup and OCT nerve, OCT macula, and FTF examination on the same day. Main Outcome Measures: Percent agreement, κ values, sensitivity, and specificity were calculated for nonexpert readers after OCT interpretation of the TECS protocol using the FTF examination as the clinical gold standard. Results: OCT did not improve the diagnostic accuracy of the TECS protocol when compared with an FTF examination. In most cases, OCT had no impact, and in the case of reader 2, OCT actually reduced the κvalue from moderate agreement to agreement equal to chance while lowering the percent agreement by 10%. OCT also did not impact inter- or intrareader variability parameters. Conclusions: In this study, OCT did not seem to improve the accuracy of glaucoma or retinal disease detection when added to the standard TECS protocol. In one case, OCT worsened the agreement of the reader compared with the FTF. Further study is necessary to confirm these findings, and results may change if the readers are glaucoma or retina specialists instead of nonexpert OCT readers, comprehensive and anterior segment specialists.

Odden, Jamie L et al (2020)[Evaluation Study] Telemedicine in long-term care of glaucoma patients [27]

Introduction This manuscript describes data from an original study, simulating a tele-glaucoma programme in an established clinic practice with an interdisciplinary team. This is a ‘real life’ trial of a telemedicine approach to see a follow-up patient. The goal is to evaluate the accuracy of such a programme to detect worsening and/or unstable disease. Such a programme is attractive since in-clinic time could be reduced for both the patient and provider. This study evaluates agreement between in-person and remote assessment of glaucoma progression. MethodsA total of 200 adult glaucoma patients were enrolled at a single institution. The in-person assessment by an optometrist or glaucoma specialist at time of enrolment was used as the gold standard for defining progression. Collated clinical data were then reviewed by four masked providers who classified glaucoma as progression or non-progression in each eye by comparing data from enrolment visit to data from the visit immediately prior to enrolment. Agreement of glaucoma progression between the masked observer and the in-person assessment was determined using Kappa statistics. Intraobserver agreement was calculated using Kappa to compare in-person to remote assessment when both assessments were performed by the same provider (n=279 eyes). Results A total of 399 eyes in 200 subjects were analysed. Agreement between in-person versus remote assessment for the determination of glaucoma progression was 63%, 62%, 69% and 68% for each reader 1–4 (kappa values=0.19, 0.20, 0.35 and 0.33, respectively). For intra-observer agreement, reader 1 agreed with their own in-person assessment for 65% of visits (kappa=0.18). Discussion Intra-observer agreement was similar to the agreement for each provider who did not see the patient in person. This similarity suggests thattelemedicine may be equally effective at identifying glaucomatous disease progression, regardless of whether the same provider performed both in-clinic and remote assessments. However, fair agreement levels highlight a limitation of using only telemedicine data to determine progression compared with clinical detail available during in-patient assessment.

Pueyo, Victoria et al (2020) [Evaluation Study] Development of a system based on artificial intelligence to identify visual problems in children: study protocol of the TrackAI project [28]

Introduction: Around 70% to 80% of the 19 million visually disabled children in the world are due to a preventable or curable disease, if detected early enough.Vision screening in childhood is an evidence-based and costeffective way to detect visual disorders. However, current screening programmes face several limitations: training required to perform them efficiently, lack of accurate screening tools and poor collaboration from young children. Some of these limitations can be overcome by new digital tools. Implementing a system based on artificial intelligence systems avoid the challenge of interpreting visual outcomes. The objective of the TrackAI Project is to develop a system to identify children with visual disorders. The system will have two main components: a novel visual test implemented in a digital device, DIVE (Device for an Integral Visual Examination); and artificial intelligence algorithms that will run on a smartphone to analyse automatically the visual data gathered by DIVE. Methods and Analysis: This is a multicentre study, with at least five centres located in five geographically diverse study sites participating in the recruitment, covering Europe, USA and Asia.The study will include children aged between 6 months and 14 years, both with normal or abnormal visual development. The project will be divided in two consecutive phases: design and training of an artificial intelligence (AI) algorithm to identify visual problems, and system development and validation. The study protocol will consist of a comprehensive ophthalmological examination, performed by an experienced paediatric ophthalmologist, and an exam of the visual function using a DIVE.For the first part of the study, diagnostic labels will be given to each DIVE exam to train the neural network. For the validation, diagnosis provided by ophthalmologists will be compared with AI system outcomes. Ethics and Dissemination: The study will be conducted in accordance with the principles of Good Clinical Practice. This protocol was approved by the Clinical Research Ethics Committee of Aragón, CEICA, on January 2019 [Code PI18/346]. Results will be published in peer-reviewed journals and disseminated in scientific meetings.

Strul, Sasha et al (2020)[Evaluation Study] Pediatric diabetic retinopathy telescreening [29]

Purpose: To describe the role of telemedicine screening for pediatric diabetic retinopathy (DR) and to identify risk factors for pediatric DR. Methods: The medical records of a telemedicine program at a tertiary, academic medical center over 17 months were reviewed retrospectively. Patients visiting an academic pediatric endocrinology clinic who met guidelines underwent telescreening. Presence of pediatric DR and risk factors for retinopathy were evaluated. Results: The fundus photographs of 852 patients 10-23 years of age were reviewed. Diabetic retinopathy was noted in 51 (6%). Patients with an abnormal screening photograph were compared to patients with diabetes who had normal screening photographs (n = 64). Older age, longer diabetes duration, type 1 diabetes, and higher average glycated hemoglobin (HbA1c)from the year prior to the photograph were associated with increased risk of retinopathy. Of these, longer duration (P = 0.003) and higher average A1c (P = 0.02) were significant after adjusting for sex, race, and age. Conclusions: Our telemedicine program found a higher percentage of diabetic retinopathy screening non-mydriatic photographs than prior studies found through standard ophthalmic examinations. In this relatively small sample size, longer duration of disease and higher average A1c were associated with increased risk of having diabetic retinopathy in our study.

Tan, Nicholas Y. Q et al (2020)[Literature Review] Glaucoma screening: where are we and where do we need to go? [30]

Purpose: Current recommendations for glaucoma screening are decidedly neutral. No studies have yet documented improved long-term outcomes for individuals who undergo glaucoma screening versus those who do not. Given the long duration that would be required to detect a benefit, future studies that may answer this question definitively are unlikely. Nevertheless, advances in artificial intelligence and telemedicine will lead to more effective screening at lower cost. With these new technologies, additional research is needed to determine the costs and benefits of screening for glaucoma. Recent Findings: Using optic disc photographs and/or optical coherence tomography, deep learning systems appear capable of diagnosing glaucoma more accurately than human graders. Eliminating the need for expert graders along with better technologies for remote imaging of the ocular fundus will allow for less expensive screening, which could enable screening of individuals with otherwise limited healthcare access. In India and China, where most glaucoma remains undiagnosed, glaucoma screening was recently found to be cost-effective. Summary: Recent advances in artificial intelligence and telemedicine have the potential to increase the accuracy, reduce the costs, and extend the reach of screening. Further research into implementing these technologies in glaucoma screening is required.

Wang, Linyan et al (2020)[Evaluation Study] Automated identification of malignancy in whole-slide pathological images: identification of eyelid malignant melanoma in gigapixel pathological slides using deep learning [3

Background/aims: To develop a deep learning system (DLS)that can automatically detect malignant melanoma (MM)in the eyelid from histopathological sections with colossal information density. Methods: Setting: Double institutional study. Study Population: We retrospectively reviewed 225 230 pathological patches [small section cut from pathologistlabelled area from an HandE image], cut from 155 HandE-stained whole-slide images (WSI). Observation Procedures: Labelled gigapixel pathological WSIs were used to train and test a model designed to assign patch-level classification. Using malignant probability from a convolutional neural network, the patches were embedded back into each WSI to generate a visualisation heatmap and leveraged a random forest model to establish a WSI-level diagnosis. Main Outcome Measure(s): For classification, the area under the receiver operating characteristic curve (AUC), accuracy, sensitivity and specificity were used to evaluate the efficacy of the DLS in detecting MM. Results: For patch diagnosis, the model achieved an AUC of 0.989 (95% CI 0.989 to 0.991), with an accuracy, sensitivity and specificity of 94.9%, 94.7% and 95.3%, respectively. We displayed the lesion area on the WSIs as graded by malignant potential. For WSI, the obtained sensitivity, specificity and accuracy were 100%, 96.5% and 98.2%, respectively, with an AUC of 0.998 (95% CI 0.994 to 1.000). Conclusion: Our DLS, which uses artificial intelligence, can automatically detect MM in histopathological slides and highlight the lesion area on WSIs using a probabilistic heatmap. In addition, our approach has the potential to be applied to the histopathological sections of other tumour types.

Wang, Sophia Y et al (2020)[Feasibility Study] Automated extraction of ophthalmic surgery outcomes from the electronic health record [32]

Objective: Comprehensive analysis of ophthalmic surgical outcomes is often restricted by limited methodologies for efficiently and accurately extracting clinical information from electronic health record (EHR) systems because much is in free-text form. This study aims to utilize advanced methods to automate extraction of clinical concepts from the EHR free text to study visual acuity (VA), intraocular pressure (IOP), and medication outcomes of cataract and glaucoma surgeries. Methods: Patients who underwent cataract or glaucoma surgery at an academic medical center between 2009 and 2018 were identified by Current Procedural Terminology codes. Rulebased algorithms were developed and used on EHR clinical narrative text to extract intraocular lens (IOL) power and implanttype, as well as to create a surgery laterality classifier. MedEx [version 1.3.7] was used on free-text clinical notes to extract information on eye medications and compared to information from medication orders. Random samples of free-text notes were reviewed by two independent masked annotators to assess interannotator agreement on outcome variable classification and accuracy of classifiers. VA and IOP were available from semi-structured fields. Results: This study cohort included 6347 unique patients, with 8550 stand-alone cataract surgeries, 451 combined cataract/glaucoma surgeries, and 961 glaucoma surgeries without concurrent cataract surgery. The rule-based laterality classifier achieved 100% accuracy compared to manual review of a sample of operative notes by independent masked annotators. For cataract surgery alone, glaucoma surgery alone, or combined cataract/glaucoma surgeries, our automated extraction algorithm achieved 99-100% accuracy compared to manual annotation of samples of notes from each group, including IOL model and IOL power for cataract surgeries, and glaucoma implant for glaucoma surgeries. For glaucoma medications, there was 90.7% inter-annotator agreement. After adjudication, 85.0% of medications identified by MedEx determined to be correct. Determination of surgical laterality enabled evaluation of pre- and postoperative VA and IOP for operative eyes. Conclusion: This text-processing pipeline can accurately capture surgical laterality and implant model usage from free-text operative notes of cataract and glaucoma surgeries, enabling extraction of clinical outcomes including visual acuities, intraocular pressure, and medications from the EHR system. Use of this approach with EHRs to assess ophthalmic surgical outcomes can benefit research groups interested in studying the safety and clinical efficacies of different surgical approaches.

Baxter, Sally L et al(2019)[Evaluation Study]Machine Learning-Based Predictive Modeling of Surgical Intervention in Glaucoma Using Systemic Data From Electronic Health Records [33]

Purpose: To predict the need for surgical intervention in patients with primary open-angle glaucoma (POAG) using systemic data in electronic health records (EHRs). Design: Development and evaluation of machine learning models. Methods: Structured EHR data of 385 POAG patients from a single academic institution were incorporated into models using multivariable logistic regression, random forests, and artificial neural networks. Leave-one-out cross-validation was performed. Mean area under the receiver operating characteristic curve (AUC), sensitivity, specificity, accuracy, and Youden index were calculated for each model to evaluate performance. Systemic variables driving predictions were identified and interpreted. Results: Multivariable logistic regression was most effective at discriminating patients with progressive disease requiring surgery, with an AUC of 0.67. Higher mean systolic blood pressure was associated with significantly increased odds of needing glaucoma surgery (odds ratio [OR] = 1.09, P < .001). Ophthalmic medications (OR = 0.28, P < .001), non-opioid analgesic medications (OR = 0.21, P = .002), anti-hyperlipidemic medications (OR = 0.39, P = .004), macrolide antibiotics (OR = 0.40, P = .03), and calcium blockers (OR = 0.43, P = .03) were associated with decreased odds of needing glaucoma surgery. Conclusions: Existing systemic data in the EHR has some predictive value in identifying POAG patients at risk of progression to surgical intervention, even in the absence of eye-specific data. Blood pressurerelated metrics and certain medication classes emerged as predictors of glaucoma progression. This approach provides an opportunity for future development of automated risk prediction within the EHR based on systemic data to assist with clinical decision-making.

Bodnar, Zachary M et al (2019)[Conference Abstract and Literature
Review] Evaluating New Ophthalmic Digital Devices for Safety and
Effectiveness in the Context of Rapid Technological Development [34]

Importance: The US Food and Drug Administration’s medical device regulatory pathway was initially conceived with hardware devices in mind. The emerging market for ophthalmic digital devices necessitates an evolution of this paradigm. Objectives: To facilitate innovation in ophthalmic digital health with attention to safety and effectiveness. Evidence Review: This article presents a summary of the presentations, discussions, and literature review that occurred during a joint Ophthalmic Digital Health workshop of the American Academy of Ophthalmology, the American Academy of Pediatrics, the American Association for Pediatric Ophthalmology and Strabismus, the American Society of Cataract and Refractive Surgery, the American Society of Retina Specialists, the Byers Eye Institute at Stanford and the US Food and Drug Administration. Findings: Criterion standards and expert graders are critically important in the evaluation of automated systems and telemedicine platforms. Training at all levels is important for the safe and effective operation of digital health devices. The risks associated with automation are substantially increased in rapidly progressive diseases. Cybersecurity and patient privacy warrant meticulous attention. Conclusions and Relevance: With appropriate attention to safety and effectiveness, digital health technology could improve screening and treatment of ophthalmic diseases and improve access to care.

Inomata, Takenori et al (2019)[Evaluation Study] Characteristics and Risk Factors Associated With Diagnosed and Undiagnosed Symptomatic Dry Eye Using a Smartphone Application [35]

Importance: The incidence of dry eye disease has increased; the potential for crowdsource data to help identify undiagnosed dry eye in symptomatic individuals remains unknown. Objective: To assess the characteristics and risk factors associated with diagnosed and undiagnosed symptomatic dry eye using the smartphone app DryEyeRhythm. Design, Setting, and Participants: A cross-sectional study using crowdsourced data was conducted including individuals in Japan who downloaded DryEyeRhythm and completed the entire questionnaire; duplicate users were excluded. DryEyeRhythm was released on November 2, 2016; the study was conducted from November 2, 2016, to January 12, 2018. Exposures: DryEyeRhythm data were collected on demographics, medical history, lifestyle, subjective symptoms, and disease-specific symptoms, using the Ocular Surface Disease Index (100-point scale; scores 0-12 indicate normal, healthy eyes; 13- 22, mild dry eye; 23-32, moderate dry eye; 33-100, severe dry eye symptoms), and the Zung Self-Rating Depression Scale (total of 20 items, total score ranging from 20-80, with ≥40 highly suggestive of depression). Main Outcomes and Measures: Multivariate-adjusted logistic regression analysis was used to identify risk factors for symptomatic dry eye and to identify risk factors for undiagnosed symptomatic dry eye. Results: A total of 21 394 records were identified in our database; 4454 users, included 899 participants (27.3%) with diagnosed and 2395 participants (72.7%) with undiagnosed symptomatic dry eye, completed all questionnaires and their data were analyzed. A total of 2972 participants (66.7%) were women; mean (SD) age was 27.9 (12.6) years. The identified risk factors for symptomatic vs no symptomatic dry eye included younger age (odds ratio [OR], 0.99; 95% CI, 0.987-0.999, P = .02), female sex (OR, 1.99; 95% CI, 1.61-2.46; P < .001), pollinosis (termed hay fever on the questionnaire)(OR, 1.35; 95% CI, 1.18-1.55; P < .001), depression (OR, 1.78; 95% CI, 1.18-2.69; P = .006), mental illnesses other than depression or schizophrenia (OR, 1.87; 95% CI, 1.24-2.82; P = .003), current contact lens use (OR, 1.27; 95% CI, 1.09-1.48; P = .002), extended screen exposure (OR, 1.55; 95% CI, 1.25-1.91; P < .001), and smoking (OR, 1.65; 95% CI, 1.37-1.98; P < .001). The risk factors for undiagnosed vs diagnosed symptomatic dry eye included younger age (OR, 0.96; 95% CI, 0.95-0.97; P < .001), male sex (OR, 0.55; 95% CI, 0.42-0.72; P < .001), as well as absence of collagen disease (OR, 95% CI, 0.23; 0.09-0.60; P = .003), mental illnesses other than depression or schizophrenia (OR, 0.50; 95% CI, 0.36-0.69; P < .001), ophthalmic surgery other than cataract surgery and laser-assisted in situ keratomileusis (OR, 0.41; 95% CI, 0.27-0.64; P < .001), and current (OR, 0.64; 95% CI, 0.54-0.77; P < .001) or past (OR, 0.45; 95% CI, 0.34-0.58; P < .001) contact lens use. Conclusions and Relevance: This study’s findings suggest that crowdsourced research identified individuals with diagnosed and undiagnosed symptomatic dry eye and the associated risk factors. These findings could play a role in earlier prevention or more effective interventions for dry eye disease.

Moss, Heather E et al (2019)[Feasibility Study] Big Data Research in Neuro-Ophthalmology: Promises and Pitfalls [36]

Background: Big data clinical research involves application of large data sets to the study of disease. It is of interest to neuro-ophthalmologists but also may be a challenge because of the relative rarity of many of the diseases treated. Evidence Acquisition: Evidence for this review was gathered from the authors’ experiences performing analysis of large data sets and review of the literature. Results: Big data sets are heterogeneous, and include prospective surveys, medical administrative and claims data and registries compiled from medical records. High-quality studies must pay careful attention to aspects of data set selection, including potential bias, and data management issues, such as missing data, variable definition, and statistical modeling to generate appropriate conclusions. There are many studies of neuro-ophthalmic diseases that use big data approaches. Conclusions: Big data clinical research studies complement other research methodologies to advance our understanding of human disease. A rigorous and careful approach to data set selection, data management, data analysis, and data interpretation characterizes high-quality studies.

Patel, Tapan P et al (2019)[Feasibility Study] Smartphone-based fundus photography for screening of plus-disease retinopathy of prematurity [37]

Background: Inadequate screening of treatment-warranted retinopathy of prematurity (ROP) can lead to devastating visual outcomes. Especially in resource-poor communities, the use of an affordable, portable, and easy to use smartphone-based non-contact fundus photography device may prove useful for screening for high-risk ROP. This study evaluates the feasibility of screening for high-risk ROP using a novel smartphone-based fundus photography device, RetinaScope. Methods: Retinal images were obtained using RetinaScope on a cohort of prematurely born infants during routine examinations for ROP. Images were reviewed by two masked graders who determined the image quality, the presence or absence of plus disease, and whether there was retinopathy that met predefined criteria for referral. The agreement between image-based assessments was compared to the gold standard indirect ophthalmoscopic assessment. Results: Fifty-four eyes of 27 infants were included. A wide-field fundus photograph was obtained using RetinaScope. Image quality was acceptable or excellent in 98% and 95% of cases. There was substantial agreement between the gold standard and photographic assessment of presence or absence of plus disease (Cohen’s κ = 0.85). Intergrader agreement on the presence of any retinopathy in photographs was also high (κ = 0.92). Conclusions: RetinaScope can capture digital retinal photographs of prematurely born infants with good image quality for grading of plus disease.

Simkin, Samantha K et al (2019)[Evaluation Study] Auckland regional telemedicine retinopathy of prematurity screening network: A 10-year review [38]

Importance: Retinopathy of prematurity (ROP)is a potentially blinding condition affecting the retinae of premature infants. Effective screening is necessary for timely treatment. Background: The Auckland Regional Telemedicine ROP (ART-ROP) network, utilizes wide-field digital imaging for ROP screening. This study reviews the ART-ROP network. Design: Retrospective analysis of the ART-ROP database. Participants: Files of infants in ART-ROP from 2006 to 2015. Methods: Data on infant demographics, ROP stage, treatment and outcome was collected. Main Outcome Measures: The efficacy of ART-ROP in the management of ROP. Results: A review of 1181 infants across three neonatal intensive care units, was completed. Infants had a mean of four screening sessions with no infants who met ROP screening criteria being missed. Type 1 ROP was present in 83 infants, who had significantly lower average birth weight 786 ± 191 g compared to 1077 ± 285 g (P < .001), and gestational age 25.3 ± 1.7 weeks compared to 27.8 ± 2.2 weeks (P < .001)than the screened cohort. The number of infants requiring screening increased (R 2 = .7993), yet treatment rates decreased (R 2= .9205) across the time period. Out-patient clinic followup was attended by 75.10% of infants screened and there was no missed ROP in those infants seen. Conclusions and Relevance:ART-ROP solely uses wide-field digital imaging for ROP diagnosis, and management, including discharge, of infants. This detailed review of ART-ROP indicates an increase in screening demand, but a decrease in the rate of type 1 ROP. The ART-ROP telemedicine model demonstrates real potential to address workforce shortage in ROP screening.

Starr, Matthew R et al (2019)[Evaluation Study] Telemedicine in the Management of Exudative Age-Related Macular Degeneration within an Integrated health care System [39]

Purpose: To investigate the outcomes of patients with exudative age-related macular degeneration (AMD)treated with intravitreal antivascular endothelial growth factors (VEGF) using a telemedicine system. Design: Interventional case series. Methods: This study examined all patients with exudative AMD who were receiving intravitreal anti-VEGF injections from September 1, 2015, through August 31, 2017, using electronic consultations at a single academic center and health system. Patients were managed initially by a retinal specialist and then allowed to receive further care with their local ophthalmologist. There were 200 electronic consultations placed during this time period for 83 eyes of 59 patients. Data collected included the retina specialist’s recommendations: intravitreal agent, interval between injections, number of injections, and when the patient was to follow-up. All occurrences of recommendations that were not completed were reported. Results: The mean age of the patients at the time of electronic consultations was 82.3 ± 7.3 years with a mean follow-up time of 2.4 ± 0.81 years. The mean distance from the home of the patient to the retina specialist was 70 ± 44 miles. There were 14 consultations (7.1%)that did not comply with the recommendations of the retina specialist. Most of these were due to other medical comorbidities leading to missed appointments or scheduling errors. Conclusions: In an integrated health care setting, 59 patients with exudative AMD were identified who were able to be effectively managed using a telemedicine system. In the appropriate setting, telemedicine may be able to assist in the management of patients with wet AMD.

Ting, Darren Shu Jeng et al (2019)[Feasibility Study] Artificial intelligence assisted telemedicine platform for cataract screening and management: a potential model of care for global eye health [40]

In their editorial, Ting et al discuss how in ophthalmology, many studies showed comparable, if not better, diagnostic performance in using artificial intelligence to screen, diagnose, predict and monitor various eye conditions on fundus photographs and optical coherence tomography, including diabetic retinopathy, age-related macular degeneration, glaucoma and retinopathy of prematurity.

Wu, Yue et al (2019)[Feasibility Study] Development and validation of a machine learning, smartphone-based tonometer [41]

Background/Aims: To compare intraocular pressure (IOP) measurements using a prototype smartphone tonometer with other tonometers used in clinical practice. Methods: Patients from an academic glaucoma practice were recruited. The smartphone tonometer uses fixed force applanation and in conjunction with a machine-learning computer algorithm is able to calculate the IOP. IOP was also measured using Goldmann applanation tonometry (GAT)in all subjects. A subset of patients were also measured using ICare, pneumotonometry [upright and supine positions] and Tono-Pen [upright and supine positions] and the results were compared. Results: 92 eyes of 81 subjects were successfully measured. The mean difference (in mm Hg)for IOP measurements of the smartphone tonometer versus other devices was +0.24 mm Hg for GAT, -1.39 mm Hg for ICare, -3.71 mm Hg for pneumotonometry and -1.30 mm Hg for Tono-Pen. The 95% limits of agreement for the smartphone tonometer versus other devices was -4.35 to 4.83 mm Hg for GAT, -6.48 to 3.70 mm Hg for ICare, -7.66 to -0.15 mm Hg for pneumotonometry and -5.72 to 3.12 mm Hg for Tono-Pen. Overall, the smartphone tonometer results correlated best with GAT (R 2=0.67, p<0.001). Of the 92 videos, 90 (97.8%) were within ±5 mm Hg of GAT and 58 (63.0%) were within ±2 mm Hg of GAT. Conclusions: Preliminary IOP measurements using a prototype smartphone-based tonometer was grossly equivalent to the reference standard.