In a Retinal Physician piece from January 2023, Miguel Busquets, MD, wrote about how artificial intelligence (AI) and augmented intelligence are weaving their way into modern ophthalmic practice.¹ Among the technologies Dr. Busquets describes is extra-clinical monitoring for conversion of intermediate age-related macular degeneration (AMD) to neovascular AMD via the ForeseeHome AMD Monitoring Program, a home-based service by the Notal Vision Monitoring Center. This article will probe deeper into home-based monitoring for conversion of intermediate AMD to neovascular AMD, in part to define the technology and in part to contextualize its utility in our patients’ ever-evolving technology-based lives.

HOME MONITORING FOR NEOVASCULAR AMD CONVERSION

Discussions about the distinction between intermediate AMD and neovascular AMD rely on the Beckman classification system as articulated by Ferris et al.² In the Beckman system, patients with at least 1 drusen at least 125 µm in size and/or any pigmentary abnormalities have intermediate AMD, whereas those with evidence of fluid leakage secondary to retinal neovascularization have neovascular AMD.

Patients with neovascular AMD are advised to routinely visit clinics for evaluation and/or therapy. Although we know that real-world patients are often lost to follow-up for various reasons, neovascular AMD patients are nevertheless within the care orbit of a clinic by virtue of the fact that their disease is documented and they are ostensibly undergoing in-office monitoring and treatment; even imperfect follow-up is better than no follow-up at all.

Patients with intermediate AMD, while still within the orbit of an eyecare practice, are less tethered to clinical care than their counterparts who receive scheduled intravitreal injections. Almost all patients with intermediate AMD are advised to return to the clinic for close follow-up examination, chiefly to detect if they have converted to neovascular AMD. In the interim, patients are sometimes advised to use an Amsler grid for self-evaluation and instructed to contact the clinic if they detect vision disruptions.

Such an analog approach presents a range of issues, including noncompliance with Amsler grid use and lack of follow-up from a patient following a visual disruption. Such a strategy is reactive (ie, responding to visual symptoms) rather than proactive (ie, acting before vision loss is noticed). Further, it presents opportunities for snags that may ultimately lead to patient dropout despite a patient’s best intentions: consider a patient who detects visual disruption on the Amsler grid, cannot find the contact information for their eye care provider, and ultimately lets follow-up slip off their radar until visual disruption becomes much worse. A recent meta-analysis study published in JAMA Ophthalmology found the Amsler grid to be inaccurate in detecting wet AMD. When performed under supervision and in the best possible conditions, the Amsler grid was positive in 2 of 3 cases with wet AMD. In patients at risk of wet AMD, the Amsler grid was positive in 1 of 3.³

Concerns about early disease detection and patient dropout among those with intermediate AMD may be alleviated by prescribing the ForeseeHome AMD Monitoring Program, which can be used alongside regular provider visits to detect conversion from intermediate AMD to neovascular AMD. To the patient, this means placing a small device in their home and performing short tests several times per week, at the direction of their prescribing clinician.

To the physician, however, an AMD monitoring program is something greater than a piece of hardware. It represents the collaboration among the provider who prescribes a home monitoring regimen, an AI-assisted algorithm that detects variations in testing outcomes worthy of further review, and, in the case of the ForeseeHome program, the on-staff ophthalmologists at the Notal Vision Monitoring Center who evaluate the tests that have been flagged for closer scrutiny. The prescribing practice is then alerted that a significant change in baseline testing has occurred, so the patient’s eye care provider can determine the appropriate next steps. This typically includes scheduling a follow-up examination as soon as possible.

Now that we know what the ForeseeHome AMD Monitoring Program is, we ought to explore what it is not. Providers who prescribe this monitoring program are not tasked with device setup, troubleshooting, and patient orientation of the device; the Notal Vision Monitoring Center team manages all of this. It does not require providers to analyze patient data each day; to do so would drain to a practice’s time, ultimately eroding the efficiencies that this technology creates. The service does not cost the practice anything; patients pay a monthly monitoring fee that is covered by Medicare.

Retina specialists have expressed a worry that embracing an at-home monitoring program that leverages artificial intelligence will lead to machines empowered to make clinical judgement calls. However, the ForeseeHome AMD Monitoring Program relies on physician judgement as much as routine clinical practice does. Artificial intelligence isn’t telling doctors that a patient needs an injection or has neovascular AMD. Rather, it advises providers to shorten the interval between the next regularly scheduled in-person examination, thereby allowing a patient to begin treatment earlier in their disease course if, and only if, their provider deems it appropriate.

VALUE OF PROMPT DETECTION OF NEOVASCULAR AMD CONVERSION

Retina specialists do not need years of real-world data to tell them what their experience has already taught them, which is that many patients with neovascular AMD benefit from prompt intervention at the beginning of their disease course. Where data may inform this broad-stroke observation, however, is the degree to which intervention following early detection of neovascular AMD benefits patients.

A retrospective study of the AAO’s IRIS Registry, a database of real-world patient data that includes more than 75 million unique patients, found that patients with neovascular AMD who present before VA has been adversely affected are more likely to continue having good vision at 2 years.⁴ Patients who presented with at least 20/40 BCVA at baseline and received at least 2 anti-VEGF injections over a 2-year period maintained a mean BCVA of at least 20/40; among those with worse than 20/40 BCVA at presentation, mean BCVA never rose above 20/40. These real-world data testify to the importance of early detection and intervention.

Data show that the ForeseeHome AMD Monitoring Program is an effective tool in helping to detect conversion from intermediate AMD to neovascular AMD in contemporary clinical settings. The ALOFT study found that, among ForeseeHome users who converted from intermediate AMD to neovascular AMD, 52% triggered a system alert prior to diagnosis of neovascular AMD, with the remaining 48% having their conversion detected following a routine or symptom-initiated visit.⁵ Importantly, these data show that the ForeseeHome AMD Monitoring Program is best used as an adjunct to, not a replacement for, routine in-person monitoring and examination. In about half of the cases, ForeseeHome prompts an office visit before the patient notices any changes in vision and would be at risk of losing further vision before diagnosis if such a safety net was not in place.

THE NEXT GENERATION OF RETINA CARE

The ForeseeHome AMD Monitoring Program is hardly the final digital innovation in retinal care. Our patients can expect to see more integration of AI-based technologies into care frameworks that increasingly rely on augmented intelligence to elevate the quality of care provided to patients.

Artificial intelligence–enabled home OCT technology may soon expand the capacity for home monitoring. Whether such technology will be adopted is a separate issue. Interestingly, data from both patients and providers suggest that there is a disconnect between what ahysicians think patients can handle and what patients can actually manage. The 2022 ASRS Preferences & Trends Survey found that 40% of US retina specialists feared that access and/or technology issues would limit the real-world use of home OCT.⁶ However, a 2022 study by Liu et al found that 93% of scans captured by a home OCT platform were able to be analyzed.⁷ In this instance, it appears that patients’ ability to effectively integrate technology into care may be underestimated by providers.

Aside from patient-facing technology, AI has been leveraged to curate and refine patient databases such as the IRIS Registry. This allows researchers and innovators to quickly search through validated registries housing real-world patient data, filtering for specific parameters and expanding the breadth of real-world patient behavior. It may also allow the identification of baseline parameters that, for example, predict responses to particular therapeutic strategies. A subanalysis of the ALOFT study highlights the predictive value of nonexudative ForeseeHome alerts for future conversion to wet AMD: patients are nearly twice as likely to convert as those without this type of alert.⁸

Retina specialists who embrace these advances will free themselves to increase face time with patients, provide earlier interventions, and improve their overall quality of care. Those who reject innovations risk grasping to retain a medical framework moving forward without them. Retina specialists should embrace AI frameworks the way we have embraced other innovations in retina care. Our patients’ vision may depend on it. RP

REFERENCES

1. Busquets M. Artificial intelligence and augmented intelligence in ophthalmic practice. Retin Phys. 2023;20(1):26-28. https://www.retinalphysician.com/issues/2023/january-february-2023/artificial-intelligence-and-augmented-intelligence
2. Ferris F, Wilkinson CP, Bird A, et al. Clinical classification of age-related macular degeneration. Ophthalmology. 2013;120(4):844-851.
3. Bjerager J, Schneider M, Potapenko I, et al. Diagnostic accuracy of the Amsler grid test for detecting neovascular age-related macular degeneration: a systematic review and meta-analysis. JAMA Ophthalmol. Published online February 16, 2023.
4. Ho AC, Kleinman DM, Lum FC, et al. Baseline visual acuity at wet AMD diagnosis predicts long-term vision outcomes: an analysis of the IRIS registry. Ophthalmic Surg Lasers Imaging Retina. 2020;51(11):633-639.
5. Mathai M, Reddy S, Elman MJ, et al. Analysis of the long-term visual outcomes of ForeseeHome remote telemonitoring: the ALOFT study. Ophthalmol Retina. 2022;6(10):922-929.
6. Stone TW, Hahn P, Raef S, eds. ASRS 2022 Preferences and Trends membership survey. Chicago, IL. American Society of Retina Specialists; 2022.
7. Liu Y, Holekamp NM, Heier JS. Prospective, longitudinal study: daily self-imaging with home OCT for neovascular age-related macular degeneration. Ophthalmol Retina. 2022;6(7):575-585.
8. Ho AC, Schechet SA, Mathai M, et al. The predictive value of false-positive ForeseeHome alerts in the ALOFT study. Ophthalmol Retina. 2023;7(2):196-198. doi:10.1016/j.oret.2022.10.009