Secondary Logo

Journal Logo

Review Article

Digital Technology for AMD Management in the Post-COVID-19 New Normal

Sim, Shaun Sebastian MD∗,†,‡; Yip, Michelle YT MD∗,†,‡; Wang, Zhaoran BSc∗,†,‡; Tan, Anna Cheng Sim MD∗,†,‡; Tan, Gavin Siew Wei MD, PhD∗,†,‡; Cheung, Chui Ming Gemmy MD, FRCOphth∗,†,‡; Chakravarthy, Usha MD, PhD§; Wong, Tien Yin MD, PhD∗,†,‡; Teo, Kelvin Yi Chong MD∗,†,‡; Ting, Daniel SW MD, PhD∗,†,‡

Author Information
Asia-Pacific Journal of Ophthalmology: January-February 2021 - Volume 10 - Issue 1 - p 39-48
doi: 10.1097/APO.0000000000000363
  • Open

Abstract

Age-related macular degeneration (AMD) is a leading cause of irreversible visual impairment in people over the age of 50 years.1 The global prevalence of any type of AMD has been reported to be 8.7%.2 Although approximately 170 million individuals are afflicted, the global prevalence of AMD is expected to increase to 288 million by 2040.2–4 AMD can be classified into early-stage, intermediate-stage, and advanced-stage AMD. Early- or intermediate-stage AMD is characterized by the presence of drusen and retinal pigment epithelial changes, and these may not affect vision significantly at these stages. Advanced-stage AMD, however, tends to be vision threatening and these individuals may present with either geographic atrophy (nonexudative AMD) or choroidal neovascularization (exudative AMD).

The management of AMD spans a large spectrum, from a simple screening of early AMD to complex, repeated treatments with intravitreal vascular endothelial growth factor (VEGF) inhibitors that require close monitoring and frequent visits in neovascular AMD (nAMD). Current management strategies in nAMD have evolved from the fixed monthly regimens proposed by the initial registration trials (CATT, MARINA, ANCHOR) to regimens such as treat and extend AMD (TREX-AMD).5–9 These regimens that take into account real-world clinical considerations, attempt to mitigate high treatment burdens but still require frequent interactions between patients and doctors in hospital settings for investigations and treatments. These management strategies have worked well to date to prevent vision loss in patients with AMD but need to adapt to face the potentially life-threatening challenges from a pandemic such as the coronavirus disease 2019 (COVID-19).

The COVID-19 was declared by the World Health Organization (WHO) as a pandemic on March 11, 2020.10–13 Ophthalmologists, like other healthcare providers, are faced with the responsibility of providing essential care while ensuring they are not vectors of the disease. In response to the crisis, many international ophthalmology societies and colleges quickly published guidelines and recommendations for best practices.14–16 Earlier this year, these ophthalmology societies recommended that ophthalmologists cease to provide any treatment other than urgent or emergent care. Several measures have been proposed to minimize COVID-19 spread among ophthalmic care providers and patients: administrative control to lower patient attendance and suspension of elective services, patient triage system at entrances to identify at-risk patients, and the promotion of the use of personal protective equipment.17–19 Although these measures were important to reduce the risk of COVID-19 spread, these measures are unsustainable in the long-term especially because we have yet to see an “end” to this pandemic.

Hence, in order to cope with these measures and the high burden of care associated with the management of nAMD, retina subspecialty clinics have been forced to adopt practices that differed from traditional treatment patterns.20 Although many of these strategies are practical and safe, there is an opportunity to further improve the management of AMD through the use of digital technology. Here, we review the current and future options of digital technology and how these may improve the management of AMD in the post-COVID world.

IMPACT OF COVID-19 WORLDWIDE

The COVID-19 pandemic that has plagued 2020 has affected people from all corners of the globe. As of December 29, 2020, over 79 million individuals globally have been infected with the novel coronavirus and over 1.7 million people people have lost their lives to it.21 The sudden surge of patients has strained healthcare infrastructure across the world, with resources stretched to the breaking point while the mortality from COVID-19 has risen steadily.22 With global recognition of the need to suppress spread, travel restrictions and social distancing were mandated by most countries in the world.23 This lock-down of movement has triggered widespread acceptance and adoption of digital communication with the intention of mitigating the social, economic, and political impact of the pandemic.24–26

DIGITAL TECHNOLOGY DURING THE COVID-19 PANDEMIC

Teleconferencing has become an essential daily tool for remote communication for purposes of work, social interaction, and medical consultations.27 Notably, Zoom Video Communications’ stock price rose dramatically with the implementation of social distancing measures, from US$68.04 on 31 December 2019 to $259.51 on 26 June 2020 on the US stock market.13 In addition, the e-commerce giants’ growth during this pandemic is proving to be substantial and multifactorial. Amazon's net sales have increased this year by 26% to US$75.5 billion in the first quarter of 2020 compared to US$59.7 billion in the first quarter of 2019.28 Alibaba Group's revenue for this year's first quarter amounting to a total of US$16.1 million compared to its counterpart in 2019 where US$13.9 million revenue was generated.29 This is contributed by a forced change in consumers’ spending behavior to online retail, which also generates demand for small businesses to utilize online retail as a platform to overcome financial difficulty during the current economic decline.30 Digital payments have also gained traction during this period. An analysis by Bain & Company estimated that there will be a 5% increase in digital payments compared to pre-COVID and e-commerce digital payments will increase by 1–2%.31 As quarantine orders have eased and people have ventured out of their homes, contactless payments have risen especially since initial reports from the World Health Organization carried warnings to the public that banknotes were capable of carrying and spreading the virus.32,33 Mastercard reacted by raising contactless payment limits across 29 countries.34

DIGITAL TECHNOLOGY IN MEDICINE

Virtual clinics and teleconsultation frameworks have been the means of provision of acute specialist care to rural areas.35 Among the first premises that encouraged the development of telemedicine was the recognition of the need to provide medical assistance to remotest areas. In the 1960s, along with the development of The Space Program in preparation for space missions, the National Aeronautic and Space Administration in the United States initiated the monitoring of astronauts’ health to provide medical aid if needed through remote teleconsultations.35,36 Since then, the use of teleconsultations has increased in medicine and ophthalmology. Australia has largely capitalized on teleconsultations and implemented schemes to review acute ophthalmological conditions remotely.37 These measures have resulted in improved patient convenience by cutting the need for extensive travel for medical advice and are also cost saving by reducing unnecessary acute transfers.38 Virtual clinics have reduced the workload on tertiary centers and improved the efficiency of eye clinics, particularly for glaucoma services.39 There has been increased adoption of digital platforms around the globe alongside the internet of things (IoT) that further allows for wider access to healthcare, eye care, and increased efficiency.40,41

In this global health emergency where medical resources outweigh demand, telemedicine has enabled the triaging of at-risk populations (Table 1). Teleconsultations have allowed patients’ symptoms to be evaluated for possible COVID-19 infections while reducing the risk of exposure to other patients, healthcare professionals, and the community.42,43 With the need for rapid detection of patients with COVID-19, several groups have also harnessed the automation of artificial intelligence (AI) to enable rapid automated diagnosis of COVID-19 through analysis of radiological imaging of the respiratory system.44–46 To enable monitoring and surveillance of this pandemic, public agencies have embraced digital technologies such as IoT, big data analytics, AI, and Blockchain.47 Use of digital tools utilizing GPS or Bluetooth tracking has been critical to aid contact tracing to identify exposed individuals and sever the chain of transmission.21 Easily accessible databases such as “Worldometer” show real-time live updates on the global number of positive cases, deaths, and recovered cases.48 Thermal camera setups at access points also provide fever screening areas, and when coupled with facial recognition software, allows rapid identification of COVID-19 suspect individuals in the community.49,50

TABLE 1 - Digital Techniques Used by Health Symptoms and Government to Aid in Tackling COVID-19 Pandemic
Strategy to Tackle COVID-19 Example Digital Tool Employed
Monitoring community cases Thermal cameras with facial recognition software49,50 Thermal imaging, AI, IoT, big data
Contact tracing with digital footprint90 IoT, big data
Real-time live updates of pandemic48 IoT, big data
Reducing spread Virtual clinics for remote consultations to reduce traffic into hospitals91 Videotelephony
Home monitoring of non-urgent diseases92 IoT, big data
Quarantine, remote working27 Blockchain, videotelephony
Delivery of medications47 Blockchain
Triage Teleconsultations for symptomatic individuals with systemic or respiratory symptoms42 Videotelephony, AI
Screening of high-risk individuals based on travel and contact history42 AI, big data
Home monitoring and triage of unstable patients93 IoT, big data
Diagnosis of confirmed cases Automated diagnosis of COVID-19 from chest imaging44,46 AI
Multimodal automated analysis of symptoms, exposure history, laboratory test, and imaging45 AI
Interventions and treatment of COVID-19 patients Electronic Intensive Care Unit monitoring programs42 IoT
AI indicates artificial intelligence; IoT, internet of things.
Proposed strategies.

DIGITAL TECHNOLOGY IN OPHTHALMOLOGY

Although digitalization has been more widely accepted with recent pressures, this fourth industrial revolution had already begun to transform various sectors with no exception to healthcare.51 Ophthalmology has been at the forefront of this foray and with its culture of innovation, it was quick to adopt these novel digital technologies, including virtual health, AI, and digital home-monitoring applications to aid in improving patient care.52 Telemedicine has been integral in improving screening programs of ophthalmological diseases.35 As the availability of experienced ophthalmologists may be scarce especially in developing countries, telemedicine may allow for more appropriate distribution of resources.53,54 This is through more people screened at an early stage of disease by specialty-trained nurses and graders, allowing ophthalmologists to focus their efforts toward managing difficult and severe cases.53,54 Telemedicine has also facilitated the management of more than 10,000 potentially complex cases in low- and middle-income countries via the Orbis Cyber-Sight telemedicine program.55 The portal provides a platform for ophthalmologists in developing countries to transmit patient data and images safely, and to consult with expert mentors in the field. With the increasing development of AI, automated diagnosis of ophthalmological conditions has helped to address concerns of limited resources by providing an automated way of triage common causes of vision-threatening conditions. There is a keen global interest in AI to detect diabetic retinopathy, a major worldwide cause of preventable blindness. Products of this are multiple algorithms able to detect not only diabetic retinopathy from fundus photographs with high sensitivity and specificity, but in addition, other related eye conditions such as diabetic macular edema, glaucoma, and age-related macular degeneration.56 With retinopathy of prematurity being the leading cause of childhood blindness causing lifelong morbidity, AI algorithms for retinopathy of prematurity has been developed to address concerns of reliability and accuracy of screening of at-risk babies especially in low- or middle-income countries, usually limited by inadequate equipment, training, and personnel.57 As the retina is the only organ in which direct observation of blood vessels in vivo is possible, studies have shown correlations with the health of the microvasculature of the heart and brain. Therefore, future developments in teleophthalmology are likely to provide opportunities to use retinal vascular health screening to identify cardiac, neurovascular, and systemic diseases.58,59

DIGITAL TECHNOLOGY IN AMD

AMD typically affects elderly patients who are also at risk of morbidity and mortality related to COVID-19. We envisage that with current and future digital tools, the management of AMD can be streamlined and adapted to reduce patients’ risk of exposure to COVID-19. This digital revolution can address all aspects of AMD management. First, AMD screening can be readily performed using fundus and Optical coherence tomography (OCT) imaging in the community. Next, patients with abnormalities detected can be referred to virtual AMD clinics for expert evaluation and a clinic visit only if treatment with intravitreal VEGF inhibitor therapy is required. Finally, once stability of the disease has been achieved, home-monitoring using digital applications or devices can be implemented to avoid unnecessary hospital visits.

AI for AMD Screening and Treatment

Screening

In the past few years, several deep learning systems (DLS) have been developed for detecting and classifying the severity of AMD based on color fundus photographs. Although the majority of these algorithms were built using the Age-related Eye Disease Study (AREDS) materials,60–62 others have come from population-based studies and diabetic retinopathy screening programs.56,63 Because the color fundus images in the AREDS dataset were captured as analog photographs, they were subsequently digitized. Whether the DLSs trained using digitized images will show similar analyzing capability if presented with images that were acquired using digital cameras has not yet been established; however, there are available and increasing number of DLS trained to identify AMD from digital camera images. Most Deep learning (DL) algorithms have demonstrated the potential to perform different AMD classification tasks at high accuracy and noninferiority when compared with retinal specialists or professional human graders.56,60–62 The DLSs utilizing AREDS dataset for both training and testing purposes obtained high performance. To exemplify, Grassmann et al obtained a sensitivity of 84.20% and specificity of 94.30%, and Burlina et al obtained an area under the curve of receiver operating curve (AUC) of 0.94–0.96. Ting et al confirmed these findings in multiethnic real-world datasets allowing for broader applications of this technique.56,60,61

Apart from digital fundus photographs, OCT can also play an important role in AMD screening. DL algorithms have been built to deliver automated segmentation and classification tasks using OCT images that are key for detecting the new onset of nAMD. The clinical application of DL on OCT scans was described by De Fauw et al.64 Using 14,884 three-dimensional OCT scan volumes, they built a two-stage framework by decoupling the segmentation and classification network, which provides 1 of 4 referral suggestions, that is, urgent, semi-urgent, routine, and observation only. This framework was tested for patient triage in an ophthalmology clinic based on more than 50 common diagnoses that can be derived from OCT, compared with retinal specialists and optometrists. The DL algorithm was comparable to the decision for “urgent referral” made by 2 expert retina specialists and was better in making this decision as compared to 2 other retinal specialists and 4 optometrists with an AUC of 0.99. A key advantage of this two-stage framework is that the model can be generalized to a new OCT device by retraining the segmentation stage with manually annotated slices, whereas the classification network remains unchanged. The error rate of the framework tested on Spectralis OCT scanner with adapted segmentation network was 3.4%, not significantly different from the error rate of 5.5% on the original device type.64

Lack of confidence in the feasibility of integration of these systems and the “black box” unaccountable nature of DLS garnered some resistance to adoption during initial proposals. However, the development of “heatmaps,” areas of focus of the DLS, provided clinicians and policymakers more understanding of the neural networks’ learning and decision-making. Figure 1 shows examples of heatmaps of our DLS showing eyes with advanced AMD compared to a normal fundus. This demonstrates that the DLS is able to identify pathology in the macula suggestive of AMD and thus classify it to have AMD. The cost-effectiveness of integrating AI solutions into screening programs have also been a considerable factor for policymakers determining widespread adoption. Xie et al explored the cost-effectiveness of a fully-automated, semi-automated model for diabetic retinopathy, a vision-threatening ophthalmological condition with national screening in many countries, which demonstrated reduced cost per patient per year, providing economic rationale to integrate AI into the screening program.65

FIGURE 1
FIGURE 1:
Heatmaps of our DLS demonstrating areas of focus of the algorithm determining AMD detection as depicted by the fluorescent green signals. A, Heatmap of the fundus with AMD showing focus in the macula where the drusen is identified and localized, and the optic disc. B, Heatmap of the fundus with no AMD demonstrating background detection of the optic disc with no focus of the macula that is non-pathological. AMD indicates age-related macular degeneration; DLS, deep learning system.

A recent feasibility study adapted the multimodal retinal image analysis consisting of fundus photographs, OCT, and OCT angiography scans. Although the training dataset was relatively small (75 participants), Vaghefi et al showed that by combining multiple modalities, the DLS accuracy increased from 91% to 96% in detecting intermediate AMD, compared to using OCT alone.66 With methods to mitigate the common problem of small datasets required for training of algorithms such as generating new images through the use of Generative Adversarial Networks, further studies could even show improvement in outcomes. Figure 2 shows examples of Generative Adversarial Network-created images of AMD compared to actual images taken of eyes with AMD, showing the ability for realistic images of AMD to be created for the benefit of future studies and DLS development. Another method to overcome the problem of small datasets is to adopt “live” clinical databases that will accumulate an increasing amount of datapoint and allow interactive improvement and refinement of DLS for patients that may have heterogeneous factors for genetic and environmental susceptibility. Thus, further research should note that it is crucial to use large multicenter datasets with various macular diseases and to incorporate a multimodal approach with clinical data, color fundus photographs, and OCT imaging, in order to enhance the generalizability of the AMD DL framework.

FIGURE 2
FIGURE 2:
Generative Adversarial Network (GAN) created images of AMD compared to real images of AMD. This may represent how we can circumvent the need for large databases for DLS training by creating new images of AMD. A, This shows an example of a real color photograph with the focus of an eye with AMD with focus and cropping of the macula. Drusen is noted in the image, suggestive of AMD. B, This shows an example of a virtual image created using GAN models of an eye with AMD that shows a high resemblance of an actual photograph of macula cropped image of an eye with AMD as seen in image (A). AMD indicates age-related macular degeneration; DLS, deep learning system.

Although the DLS is able to screen for AMD, there are potential limitations in its clinical use as a screening tool for some ocular conditions as it does not address comorbid conditions or risk factors before the development of ophthalmoscopic findings. For example, elevated intraocular pressure before glaucomatous optic neuropathy and elevated glycated hemoglobin before worsening diabetic retinopathy. Thus, one would be cautious to interpret DLS screening results in the isolation of these risk factors.

Treatment

In the management of nAMD, OCT monitoring of morphology of retinal lesions and disease activity forms an integral part of monitoring and decision for retreatment. With increasing demands for effective and accurate OCT monitoring, automated techniques have been explored.67,68 These include the Notal OCT Analyzer that reports high-concordance rates when comparing to retinal specialists with an accuracy of 91%, sensitivity of 92%, and specificity of 91%.68 A recent study utilizing the AREDS2 10-year Follow-On study (AREDS2-10Y) with 1127 eyes with longitudinal data showed that the AI-based algorithm achieved a higher level of performance at detecting the presence of retinal fluid than human retinal specialists (accuracy 0.851 versus 0.805, sensitivity 0.822 versus 0.468, specificity 0.865 versus 0.970, respectively) when compared to the ground truth of expert graders at the University of Wisconsin Fundus Photograph Reading Center.67

AI has lent its ability to develop further tools to aid the management of nAMD by defining the disease and detection of biomarkers of nAMD. Various algorithms have been developed to monitor nAMD using biomarkers including but not limited to intraretinal fluid, subretinal fluid, pigment epithelial detachment, drusen, and geographical atrophy.69 The development of systems to better quantify and measure retinal fluid will enable a reliable assessment of response to anti-VEGF treatment compared to the qualitative evaluation currently used in clinical practice.70 Schmidt-Erfurth et al utilized automated segmentation methods in deep learning to ascertain volumes of intraretinal fluid, subretinal fluid, and pigment epithelial detachment, and applied this algorithm to a phase III HARBOR clinical trial.70 This study demonstrated that the stricter treatment arm with a higher dose and regular monthly dosing of anti-VEGF resulted in the least residual fluid, exemplifying how these AI algorithms are enabling improving therapeutic regimes for nAMD.70

Personalized medicine has been highly regarded as the gold standard of treatment options for an individual where treatment is optimized to minimize side effects and maximize efficacy. Development of AI algorithms aiming to predict progression to late dry and late wet AMD based on color fundus photographs allows higher-risk individuals who may benefit from closer follow surveillance and may guide management by advising on better control of risk factors and for alternative advanced treatment.71,72 In the current technological climate, models have been having high-accuracy rates with detection, but accuracy in prediction of progression of AMD has not achieved such high rates yet that proves to be a difficult task with few groups attempting this.71 In addition, predicting the onset of the disease is currently difficult with limited images of asymptomatic patients. Prognostication of functional improvement from treatment through algorithms has been explored to provide predictive information that can assist in patients’ autonomous decision-making. A study utilizing machine learning to predict visual acuity (VA) after the commencement of anti-VEGF treatment at 3- and 12-month intervals showed that the difference between algorithmic prediction and actual VA was between 0.11 and 0.18 logMAR for 3-month forecast and 0.16 and 0.22 logMAR.73 This demonstrates the utility that AI may provide in the creation of personalized medicine.

Virtual AMD Clinics for Diagnosis and Treatment

The concept of “virtual” (without actual consultation) medical retina clinics emerged in 2015.74 In this “virtual” clinic,” all patients had VA tests and OCT scan performed and reviewed asynchronously by the medical team without a face to face specialist consultation. The implementation of these virtual AMD clinics not only assisted with a reduction in mean time between consecutive appointments and waiting times, but also resulted in significant visual gains.74 Furthermore, previous studies have demonstrated high inter-reader and intrareader agreement for OCT scans.75,76 In another study, a reduction in healthcare burden was demonstrated when up to 44% of patients were found to be suitable for virtual clinics where only OCT and ultra-widefield imaging were performed.77

With the advances in imaging coupled with potential AI-driven decision making, we expect that the monitoring of treatment response, recurrence during maintenance phase or after cessation of treatment, and observation of the fellow eye for incipient neovascularization can be managed using “virtual” clinics. This concept adheres to the principles of social distancing, which is imperative for the prevention of COVID-19 spread, by allowing the acquisition of images and subsequent decision making to be separated in time and space.

By far the largest survey of patients’ attitudes regarding attending virtual clinics revealed that more than 86% of patients were supportive of a “virtual” clinic review in place of face-to-face clinic appointments.78

Real-World AMD Retina Clinics

nAMD is a sight-threatening condition that requires prompt and regular intravitreal anti-VEGF injections. Retina subspecialists treating nAMD have developed various strategies during the initial “lockdown phase.” The aim was to reduce the contact time between ophthalmic care providers and patients and congestion within the clinic, while maintaining visual improvement and/or visual stability. One of the strategies we implemented (Table 2) included a variation of the “Treat-and-Plan” regime as described by Antaki et al.20

TABLE 2 - The Clinical Protocol for New Referral Visits Versus Follow-Up Assessment Visits Versus Treatment-Only Visit
Type of Visit New Referral Visit Follow-Up Assessment Visit Treatment-Only Visit
Investigation VA, OCT, DFE - If VA and symptomatically stable: VA, OCT, or DFE- If VA and symptomatically worse: VA, OCT, DFE ± FA, ICG, OCTA None
Treatment Loading dose phase – 3 monthly intravitreal injections No disease activity: allocate to fixed-interval regime based on last stable treatment intervalPresence of disease activity: decrease treatment interval by 2 weeks For intravitreal injection
DFE indicates dilated fundal examination; ICG,Indocyanine green angiography; OCT, Optical coherence tomography; OCTA, Optical coherence tomography angiography; VA, visual acuity.

This involved 3 types of visits, new referral visit, follow-up assessment visit, and treatment only visits (Table 2). In our setting, we performed similar tests (VA, dilation, and OCT) during the new referral visits, with FA, Indocyanine green angiography, and Optical coherence tomography angiography performed only upon request of the treating physician. Patients on active treatment who attended follow-up assessment visits had VA tests with either OCT or dilated fundal examination. If these patients were deemed stable by the treating physician the patients then proceeded directly for treatment. However, if vision had declined by 1 line from the previous visit or the patient reported worse vision, both OCT and dilated fundal examination were required and treatment interval titrated based on activity status. We also instituted treatment-only visits in which patients on fixed or regular intervals attended for treatment only without further tests or investigations. Although these measures help reduce contact time and congestion, a potential downside exists for stable patients due to the inability to extend treatment intervals.

The future development of a pandemic-ready, robust clinical management must address new referrals and follow-up protocols, patient treatment compliance, tracking of clinical outcomes, and also data security.

Digital monitoring devices for AMD – ForeseeHome, myVisionTrack, Alleye

Since the late 1960s, self-monitoring in AMD patients has traditionally involved the use of an Amsler chart (grid). The Amsler grid can evaluate the central 20° visual field when used at a 30-cm testing distance.79 The identification of subtle changes in visual function (such as distortion) may suggest AMD disease activity or recurrence. The limitations of the Amsler chart include its subjective and qualitative nature, crowding effects, and perceptual completion phenomenon, hence limiting its sensitivity in detecting AMD-related visual changes.80

There are several alternatives to the Amsler chart. In an effort to improve AMD disease monitoring and recurrence preferential hyperacuity perimetry (PHP) was developed by Loewenstein et al.81 The initial technique was more sensitive than the traditional Amsler chart but had a relatively high rate of false positives. Further iterations of PHP was able to differentiate recent-onset nAMD from intermediate AMD with higher sensitivity and specificity.82 Recently, portable home monitoring devices such as the ForeseeHome AMD monitor utilizes PHP testing to detect new choroidal neovascularization development at an earlier stage83 (Fig. 3).

FIGURE 3
FIGURE 3:
Home monitoring devices for AMD from Notal Vision.94 A, Notal Vision's ForeseeHome® enables patients with intermediate AMD with VA of 20/60 or worse to take daily tests that are subsequently sent to the Notal Vision Data Monitoring Center in which the physician will be notified when changes are noted from baseline. This machine utilizes a closed viewer and a mouse for patients to click to identify visual distortions displayed to the patient's eye. B, Notal Vision's Home OCT combines OCT imaging technology with artificial intelligence software Notal OCT Analyzer (NOA™) to enable home-based monitoring of exudative AMD. It is still undergoing FDA approval currently and not available for clinical use as of yet. AMD indicates age-related macular degeneration; DLS, deep learning system; VA, visual acuity.

The shape discrimination hyperacuity (SDH) testing is another method of early identification of AMD and its progression.84 Wang et al found that a mobile version of the SDH test, myVisionTrack (mVT) developed by Genentech USA, Inc, was comparable to the previously established desktop SDH. This provides patients with a new tool that is intuitive and readily accessible to monitor macular diseases at home.85 A subsequent study by the same group also showed that elderly patients were willing to comply with this novel method of self-monitoring.86

Another recently developed novel mobile application, Alleye developed by Oculocare medical Inc. in Switzerland, uses an alignment hyperacuity task (dot alignment) to monitor visual function.87 In contrast to mVT that detects and characterizes the central 3° of metamorphopsia, Alleye screens 12.7 thus covering almost the entire macular region. The extended area of screening is useful when considering macular pathology typically extends within the vascular arcades (Fig. 4). Further studies are currently underway for evaluating the reliability of different tests for monitoring disease progression and for early detection of fellow eye involvement.88,89

FIGURE 4
FIGURE 4:
Alleye phone application that enables home-based self-monitoring of central and paracentral metamorphopsia with existing mobile phone devices in an at-risk population.95 The figure shows snapshots of the application demonstrating its use. The left image shows easy-to-follow guides for standardization of the position of the mobile phone to ensure testing of the macula. The middle image shows instructions to the user and patient on the requirements to place the middle of 3 points on the invisible connecting line between the outer points utilizing controls on the screen. The right image demonstrates an example of a test of the task previously described.

Real-World Experience with Alleye Home Monitoring Application

In Singapore National Eye Center, we deployed the Alleye application to patients who had their appointments deferred over the COVID-19 pandemic lockdown in Singapore. Patients scheduled for a follow-up in the retina clinic for any condition over the lockdown period (end April 2020 to middle June 2020) were deferred based on electronic chart review by treating physicians. All deferred patients were subsequently invited to participate in a pilot test of the Alleye app with an aim to detect early changes in vision due to reactivation of disease.

CONCLUSIONS

COVID-19 was an unexpected catalyst for digitalization progression in this century and transformation of people's daily lives. Although digital strategies were employed to monitor and curb the community spread of the virus, these initiatives were also extended to address preexisting demands on healthcare. This has become part and parcel of our “new normal” in healthcare provision and has become essential in the fight against this pandemic. In the ophthalmology community, the management of AMD has been profoundly affected as these patients are not only most vulnerable to COVID-19 infections but also require regular monitoring and treatment to slow the progression of the disease. There is an urgent unmet need to transform the way care is provisioned for AMD during this crisis and beyond. Digital transformation may be the solution for ensuring safety in all aspects of AMD management. This transformation includes advanced analytic techniques such as AI to detect and screen for disease, novel models of care that ensure minimal contact and social interactions, treatment strategies such as the “Treat-and-Plan” regime, and digital home monitoring initiatives that can detect early changes of AMD. In the post-COVID-19 new normal, we may see these strategies become more prevalent as evidence of their effectiveness to provide safe care materializes. With the continued evolution and improvement of digitalization, we will be better equipped to face the next challenge.

REFERENCES

1. Pennington KL, DeAngelis MM. Epidemiology of age-related macular degeneration (AMD): associations with cardiovascular disease phenotypes and lipid factors. Eye Vis 2016; 3:34.
2. Wong WL, Su X, Li X, et al. Global prevalence of age-related macular degeneration and disease burden projection for 2020 and 2040: a systematic review and meta-analysis. Lancet Glob Health 2014; 2 (2):e106–e116.
3. Pascolini D, Mariotti SP. Global estimates of visual impairment: 2010. Br J Ophthalmol 2012; 96 (5):614–618.
4. Sources for Macular Degeneration: Facts & Figures. http://wwwbrightfocusorg/sources-macular-degeneration-facts-figures. Accessed August 2020.
5. Rosenfeld PJ, Brown DM, Heier JS, et al. Ranibizumab for neovascular age-related macular degeneration. N Engl J Med 2006; 355 (14):1419–1431.
6. Brown DM, Michels M, Kaiser PK, et al. Ranibizumab versus verteporfin photodynamic therapy for neovascular age-related macular degeneration: two-year results of the ANCHOR study. Ophthalmology 2009; 116 (1):57–65.e5.
7. Comparison of Age-related Macular Degeneration Treatments Trials Research G, Martin DF, Maguire MG, et al. Ranibizumab and bevacizumab for treatment of neovascular age-related macular degeneration: two-year results. Ophthalmology 2012; 119 (7):1388–1398.
8. Wykoff CC, Croft DE, Brown DM, et al. Prospective trial of treat-and-extend versus monthly dosing for neovascular age-related macular degeneration: TREX-AMD 1-year results. Ophthalmology 2015; 122 (12):2514–2522.
9. Rufai SR, Almuhtaseb H, Paul RM, et al. A systematic review to assess the ’treat-and-extend’ dosing regimen for neovascular age-related macular degeneration using ranibizumab. Eye 2017; 31 (9):1337–1344.
10. Zhu N, Zhang D, Wang W, et al. A novel coronavirus from patients with pneumonia in China, 2019. N Engl J Med 2020; 382 (8):727–733.
11. Sohrabi C, Alsafi Z, O’Neill N, et al. World Health Organization declares global emergency: a review of the 2019 novel coronavirus (COVID-19). Int J Surg 2020; 76:71–76.
12. World Health Organisation WHO Director-General's opening remarks at the media briefing on COVID-19. 2020. https://wwwwhoint/dg/speeches/detail/who-director-general-s-opening-remarks-at-the-media-briefing-on-covid-19---11-march-2020. Accessed March 29, 2020.
13. Nicola M, O’Neill N, Sohrabi C, Khan M, et al. Evidence based management guideline for the COVID-19 pandemic - review article. Int J Surg 2020; 77:206–216.
14. Korobelnik JF, Loewenstein A, Eldem B, et al. Guidance for anti-VEGF intravitreal injections during the COVID-19 pandemic. Graefes Arch Clin Exp Ophthalmol 2020; 258 (6):1149–1156.
15. Teo KYC, Chan RVP, Cheung CMG. Keeping our eyecare providers and patients safe during the COVID-19 pandemic. Eye 2020; 34 (7):1161–1162.
16. Lai THT, Tang EWH, Chau SKY, et al. Stepping up infection control measures in ophthalmology during the novel coronavirus outbreak: an experience from Hong Kong. Graefes Arch Clin Exp Ophthalmol 2020; 258 (5):1049–1055.
17. Wilder-Smith A, Freedman DO. Isolation, quarantine, social distancing and community containment: pivotal role for old-style public health measures in the novel coronavirus (2019-nCoV) outbreak. J Travel Med 2020; 27 (2):
18. Ahmed F, Zviedrite N, Uzicanin A. Effectiveness of workplace social distancing measures in reducing influenza transmission: a systematic review. BMC Public Health 2018; 18 (1):518.
19. Koh A, Chen Y. Perspective from Singapore and China on the COVID-19 pandemic: the new world order for ophthalmic practice. Ophthalmology 2020; 127 (8):e49–e50.
20. Antaki F, Dirani A. Treating neovascular age-related macular degeneration in the era of COVID-19. Graefes Arch Clin Exp Ophthalmol 2020; 258 (7):1567–1569.
21. WHO. Situation Report – 194. WHO; 2020.
22. Gates B. Responding to Covid-19—a once-in-a-century pandemic? New England Journal of Medicine 2020; 382 (18):1677–1679.
23. Organisation ICA, Organisation WH. Joint ICAO-WHO Statement on COVID-19 2020.
24. Economist T. The changes covid-19 is forcing on to business. 2020. https://www.economist.com/briefing/2020/04/11/the-changes-covid-19-is-forcing-on-to-business. Accessed July 19, 2020.
25. Whitelaw S, Mamas MA, Topol E, et al. Applications of digital technology in COVID-19 pandemic planning and response. Lancet Digit Health 2020; 2:e435–e440.
26. Baig A, Hall B, Jenkins P, et al. The COVID-19 recovery will be digital: A plan for the first 90 days. 2020. https://www.mckinsey.com/business-functions/mckinsey-digital/our-insights/the-covid-19-recovery-will-be-digital-a-plan-for-the-first-90-days#. Accessed 18 July 2020.
27. Drake Bennett NG. Zoom Goes From Conferencing App to the Pandemic's Social Network. 2020. https://www.bloomberg.com/news/features/2020-04-09/zoom-goes-from-conferencing-app-to-the-pandemic-s-social-network.
28. Amazon. Amazon.com Announces First Quarter Results. Amazon; 2020.
29. Alibaba. March quarter 2020 and full fiscal year 2020 results. Alibaba Group 2020.
30. Monica P. The way we shop has fundamentally changed. That's good news for Alibaba. CNN Business 2020.
31. Gringoli V, Williams G, Ott J, et al. The Covid-19 Tipping Point for Digital Payments. 2020. https://www.bain.com/insights/the-covid-19-tipping-point-for-digital-payments/. Accessed June 29, 2020.
32. Walden S. Banking After COVID-19: The Rise of Contactless Payments in the U.S. 2020. https://www.forbes.com/advisor/banking/banking-after-covid-19-the-rise-of-contactless-payments-in-the-u-s/. Accessed June 29, 2020.
33. Tsang W. Mastercard study shows consumers globally make the move to contactless payments for everyday purchases. Seeking Touch-Free Payment Experiences Mastercard 2020.
34. Mastercard. Mastercard enables Contactless limit raise across 29 countrise; and champions permanent increase. 2020. https://newsroom.mastercard.com/eu/press-releases/mastercard-enables-contactless-limit-raise-across-29-countries-and-champions-permanent-increase/. Accessed June 29, 2020.
35. Li HK. Telemedicine and ophthalmology. Surv Ophthalmol 1999; 44 (1):61–72.
36. Baldwin C, Simpson A. A Brief History of NASA's Contributions to Telemedicine. 2013. https://www.nasa.gov/content/a-brief-history-of-nasa-s-contributions-to-telemedicine. Accessed July 29, 2020.
37. Blackwell NA, Kelly GJ, Lenton LM. Telemedicine ophthalmology consultation in remote Queensland. Med J Aust 1997; 167 (11–12):583–586.
38. Rosengren D, Blackwell N, Kelly G, et al. The use of telemedicine to treat ophthalmological emergencies in rural Australia. J Telemed Telecare 1998; 4: (suppl 1): 97–99.
39. Clarke J, Puertas R, Kotecha A, et al. Virtual clinics in glaucoma care: face-to-face versus remote decision-making. Br J Ophthalmol 2017; 101 (7):892–895.
40. Perednia DA, Allen A. Telemedicine technology and clinical applications. JAMA 1995; 273 (6):483–488.
41. CRC Press, Wootton R, Craig J, Patterson V. Introduction to Telemedicine. 2017.
42. Hollander JE, Carr BG. Virtually perfect? Telemedicine for COVID-19. N Engl J Med 2020; 382 (18):1679–1681.
43. Economist T. Telemedicine is essential amid the covid-19 crisis and after it. 2020. https://www.economist.com/open-future/2020/03/31/telemedicine-is-essential-amid-the-covid-19-crisis-and-after-it. Accessed July 15, 2020.
44. Murphy K, Smits H, Knoops AJ, et al. COVID-19 on the chest radiograph: a multi-reader evaluation of an AI system. Radiology 2020; 201874.
45. Mei X, Lee H-C, Diao KY, et al. Artificial intelligence–enabled rapid diagnosis of patients with COVID-19. Nat Med 2020; 26:1224–1228.
46. Ghinai I, McPherson TD, Hunter JC, et al. First known person-to-person transmission of severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) in the USA. Lancet 2020; 395:1137–1144.
47. Ting DSW, Carin L, Dzau V, et al. Digital technology and COVID-19. Nat Med 2020; 26 (4):459–461.
48. Worldometer. https://www.worldometers.info/coronavirus/. Accessed June 27, 2020.
49. Imaging V. https://www.viperimaging.com. Accessed June 27, 2020.
50. Borak M. Beijing considers using facial recognition to fight a new Covid-19 outbreak. 2020. https://www.scmp.com/abacus/tech/article/3089378/beijing-considers-using-facial-recognition-fight-new-covid-19-outbreak. Accessed June 27, 2020.
51. SpringerJayanthi P, Iyyanki M, Mothkuri A, et al. Fourth industrial revolution: an impact on health care industry. International Conference on Applied Human Factors and Ergonomics. 2019; 58–69.
52. Ting DS, Lin H, Ruamviboonsuk P, et al. Artificial intelligence, the internet of things, and virtual clinics: ophthalmology at the digital translation forefront. Lancet Digit Health 2020; 2 (1):e8–e9.
53. Skalet AH, Quinn GE, Ying G-S, et al. Telemedicine screening for retinopathy of prematurity in developing countries using digital retinal images: a feasibility project. J AAPOS 2008; 12 (3):252–258.
54. Ells AL, Holmes JM, Astle WF, et al. Telemedicine approach to screening for severe retinopathy of prematurity: a pilot study. Ophthalmology 2003; 110 (11):2113–2117.
55. Cyber-Sight: Existing or expanding (post-pilot). Botswana: Center for Health Market Innovations; 2002. http://healthmarketinnovations.org/program/cyber-sight.
56. Ting DSW, Cheung CY-L, Lim G, et al. Development and validation of a deep learning system for diabetic retinopathy and related eye diseases using retinal images from multiethnic populations with diabetes. JAMA 2017; 318 (22):2211–2223.
57. Gensure RH, Chiang MF, Campbell JP. Artificial intelligence for retinopathy of prematurity. Curr Opin Ophthalmol 2020; 31 (5):312–317.
58. Wong TY, Klein R, Klein BE, et al. Retinal microvascular abnormalities and their relationship with hypertension, cardiovascular disease, and mortality. Surv Ophthalmol 2001; 46 (1):59–80.
59. Wang JJ, Liew G, Wong TY, et al. Retinal vascular calibre and the risk of coronary heart disease-related death. Heart 2006; 92 (11):1583–1587.
60. Burlina PM, Joshi N, Pekala M, et al. Automated grading of age-related macular degeneration from color fundus images using deep convolutional neural networks. JAMA Ophthalmol 2017; 135 (11):1170–1176.
61. Grassmann F, Mengelkamp J, Brandl C, et al. A deep learning algorithm for prediction of age-related eye disease study severity scale for age-related macular degeneration from color fundus photography. Ophthalmology 2018; 125 (9):1410–1420.
62. Peng Y, Dharssi S, Chen Q, et al. DeepSeeNet: a deep learning model for automated classification of patient-based age-related macular degeneration severity from color fundus photographs. Ophthalmology 2019; 126 (4):565–575.
63. Liefers B, Colijn JM, González-Gonzalo C, et al. A deep learning model for segmentation of geographic atrophy to study its long-term natural history. Ophthalmology 2020; 127:1086–1096.
64. De Fauw J, Ledsam JR, Romera-Paredes B, et al. Clinically applicable deep learning for diagnosis and referral in retinal disease. Nat Med 2018; 24 (9):1342–1350.
65. Xie Y, Nguyen QD, Hamzah H, et al. Artificial intelligence for teleophthalmology-based diabetic retinopathy screening in a national programme: an economic analysis modelling study. Lancet Digit Health 2020.
66. Vaghefi E, Hill S, Kersten HM, et al. Multimodal retinal image analysis via deep learning for the diagnosis of intermediate dry age-related macular degeneration: a feasibility study. J Ophthalmol 2020; 2020:7493419.
67. Keenan TD, Clemons TE, Domalpally A, et al. Retinal specialist versus artificial intelligence detection of retinal fluid from OCT: age-related eye disease study 2: 10-year follow-on study. Ophthalmology 2020.
68. Chakravarthy U, Goldenberg D, Young G, et al. Automated identification of lesion activity in neovascular age-related macular degeneration. Ophthalmology 2016; 123 (8):1731–1736.
69. Wintergerst MW, Schultz T, Birtel J, et al. Algorithms for the automated analysis of age-related macular degeneration biomarkers on optical coherence tomography: a systematic review. Transl Vis Sci Technol 2017; 6 (4):10.
70. Schmidt-Erfurth U, Vogl W-D, Jampol LM, et al. Application of automated quantification of fluid volumes to anti-VEGF therapy of neovascular age-related macular degeneration. Ophthalmology 2020; 127:1211–1219.
71. Bhuiyan A, Wong TY, Ting DSW, et al. Artificial intelligence to stratify severity of age-related macular degeneration (AMD) and predict risk of progression to late AMD. Transl Vis Sci Technol 2020; 9 (2):25.
72. Schmidt-Erfurth U, Waldstein SM, Klimscha S, et al. Prediction of individual disease conversion in early AMD using artificial intelligence. Invest Ophthalmol Vis Sci 2018; 59 (8):3199–3208.
73. Rohm M, Tresp V, Müller M, et al. Predicting visual acuity by using machine learning in patients treated for neovascular age-related macular degeneration. Ophthalmology 2018; 125 (7):1028–1036.
74. Tsaousis KT, Empeslidis T, Konidaris VE, et al. The concept of virtual clinics in monitoring patients with age-related macular degeneration. Acta Ophthalmol 2016; 94 (5):e353–e355.
75. Zhang N, Hoffmeyer GC, Young ES, et al. Optical coherence tomography reader agreement in neovascular age-related macular degeneration. Am J Ophthalmol 2007; 144 (1):37–44.
76. Ritter M, Elledge J, Simader C, et al. Evaluation of optical coherence tomography findings in age-related macular degeneration: a reproducibility study of two independent reading centres. Br J Ophthalmol 2011; 95 (3):381–385.
77. Lee JX, Manjunath V, Talks SJ. Expanding the role of medical retina virtual clinics using multimodal ultra-widefield and optical coherence tomography imaging. Clin Ophthalmol 2018; 12:2337–2345.
78. Ahnood D, Souriti A, Williams GS. Assessing patient acceptance of virtual clinics for diabetic retinopathy: a large scale postal survey. Can J Ophthalmol 2018; 53 (3):207–209.
79. Crossland M, Rubin G. The Amsler chart: absence of evidence is not evidence of absence. Br J Ophthalmol 2007; 91 (3):391–393.
80. Trevino R. Recent progress in macular function self-assessment. Ophthalmic Physiol Opt 2008; 28 (3):183–192.
81. Loewenstein A, Malach R, Goldstein M, et al. Replacing the Amsler grid: a new method for monitoring patients with age-related macular degeneration. Ophthalmology 2003; 110 (5):966–970.
82. Alster Y, Bressler NM, Bressler SB, et al. Preferential hyperacuity perimeter (PreView PHP) for detecting choroidal neovascularization study. Ophthalmology 2005; 112 (10):1758–1765.
83. Group AHSR, Chew EY, Clemons TE, et al. Randomized trial of a home monitoring system for early detection of choroidal neovascularization home monitoring of the eye (HOME) study. Ophthalmology 2014; 121 (2):535–544.
84. Wang YZ, Wilson E, Locke KG, et al. Shape discrimination in age-related macular degeneration. Invest Ophthalmol Vis Sci 2002; 43 (6):2055–2062.
85. Wang YZ, He YG, Mitzel G, et al. Handheld shape discrimination hyperacuity test on a mobile device for remote monitoring of visual function in maculopathy. Invest Ophthalmol Vis Sci 2013; 54 (8):5497–5505.
86. Kaiser PK, Wang YZ, He YG, et al. Feasibility of a novel remote daily monitoring system for age-related macular degeneration using mobile handheld devices: results of a pilot study. Retina 2013; 33 (9):1863–1870.
87. Schmid MK, Thiel MA, Lienhard K, et al. Reliability and diagnostic performance of a novel mobile app for hyperacuity self-monitoring in patients with age-related macular degeneration. Eye 2019; 33 (10):1584–1589.
88. Ward E, Wickens RA, O’Connell A, et al. Monitoring for neovascular age-related macular degeneration (AMD) reactivation at home: the MONARCH study. Eye (Lond) 2020.
89. https://njl-admin.nihr.ac.uk/document/download/2007200. EDNA: Early Detection of Neovascular Age-related macular degeneration protocol. 2015.
90. Hellewell J, Abbott S, Gimma A, et al. Feasibility of controlling COVID-19 outbreaks by isolation of cases and contacts. Lancet Glob Health 2020; 8:e488–e496.
    91. Saleem SM, Pasquale LR, Sidoti PA, et al. Virtual ophthalmology: telemedicine in a Covid-19 era. Am J Ophthalmol 2020; 216:237–242.
    92. Shehav-Zaltzman G, Segal G, Konvalina N, et al. Remote glucose monitoring of hospitalized, quarantined patients with diabetes and COVID-19. Diabetes Care 2020; 43 (7):e75–e76.
    93. Annis T, Pleasants S, Hultman G, et al. Rapid implementation of a COVID-19 remote patient monitoring program. J Am Med Inform Assoc 2020; 27:1326–1330.
    94. Vision N. Notal Vision Technology. 2020. https://notalvision.com/technology/home-oct. Accessed August 26, 2020.
      95. Alleye. Alleye. 2020. https://alleye.io/user. Accessed August 26, 2020.
        Keywords:

        AI; AMD; Covid-19; digital healthcare; telemedicine

        Copyright © 2021 Asia-Pacific Academy of Ophthalmology. Published by Wolters Kluwer Health, Inc. on behalf of the Asia-Pacific Academy of Ophthalmology.