Contributor: AI-Based Remote Monitoring for Age-Related Macular Degeneration: Promise, Progress, and Pitfalls

Neovascular age-related macular degeneration (nAMD) is a leading cause of irreversible vision impairment in adults, affecting an estimated 1.49 million Americans.1 In-clinic intravitreal anti-VEGF injections are the standard of care for preventing vision loss and improving vision in nAMD. Medicare spending on anti-VEGF agents exceeded $4 billion annually in 2019 and may continue to grow with the advent of new therapies.² Compared with traditional anti-VEGF regimens using bevacizumab, ranibizumab, and aflibercept 2 mg, newer agents, including faricimab and aflibercept 8 mg, may offer extended durability with treatment intervals potentially up to 16 weeks, although at higher costs that could amplify the health care system burden.^3-6 Furthermore, despite the promise of long-term treatment benefits, many patients require these higher-cost therapies as often as every 4 weeks, further accruing increased system costs. Current nAMD management consists of regular office-based optical coherence tomography (OCT), dilated retinal fundus examinations, and clinical assessment to guide anti-VEGF treatment decisions. The dominant paradigm is a “treat-and-extend” model, in which a patient who has been stabilized with no active hemorrhage or retinal fluid accumulation is treated and then has their injection interval extended by 1 to 2 weeks.⁷ This paradigm works well in practice, but as a patient’s interval is extended, it may carry the risk of overtreatment in stable patients and undertreatment in those with rapidly progressive disease.

The FDA-cleared Notal Vision system is a home OCT (hOCT) technology enabling patients with nAMD to self-acquire daily retinal images with reliable artificial intelligence (AI) detection of retinal fluid volumes associated with vision loss.^8-10 Studies have shown the device’s potential to inform physician decision-making based on fluid dynamics, supporting personalized injection intervals that may circumvent the limitations of office-based monitoring alone.^11-13 Clinical studies have found that patients requiring injections every 8 weeks with standard monitoring may achieve extended injection intervals of every 15.3 weeks with hOCT, while maintaining an adherence of 6.2 home scans per week.¹⁴ Additional studies have shown that physician scan interpretation can be abbreviated with a snapshot-like review once every 30 days, potentially easing the physician reading burden of hOCT, while allowing the AI-based interpretation to inform physicians of any acute concern.¹⁵

Although AI-backed technologies often promise improved patient care and health care cost savings, our analysis of hOCT implementation reveals a more nuanced landscape of competing adoption dynamics: new expenses from device fees, AI analysis, and physician review time vs potential savings from reduced injection frequency and fewer office visits. Economic feasibility hinges on interactions between device costs, anti-VEGF agent selection, baseline treatment burden, and achieved injection-sparing magnitude. We modeled cost trajectories across injection frequencies and agents, as well as physician reading burden, to identify conditions enabling practical implementation of hOCT for nAMD monitoring.

We measured costs of hOCT vs standard monitoring using published clinical injection intervals for each monitoring modality.^14,15 Costs incorporated January 2026 Medicare Part B payment limits for 5 anti-VEGF agents, 2026 Physician Fee Schedule fees for injection-related procedures (intravitreal injection, OCT interpretation, office examination, endophthalmitis management), and remote OCT Current Procedural Terminology (CPT) codes. The hOCT costs included technical support (CPT code 0605T); because costs have not yet been publicized for initial device provision (0604T) and physician interpretation of scans over a 30 day period (0606T), we conservatively estimated physician interpretation equivalent to a single in-office OCT reading per month, and no device acquisition cost to the payer. Physician scan interpretation burden was estimated from published patient adherence rates and time-based review thresholds,^14,15 and assuming 1 minute of review per scanned eye.

Preliminary results of our ongoing research indicate that cost savings in hOCT use may be achieved by deploying the technology in patients with high injection burdens receiving costly anti-VEGF agents, because injection-sparing benefits and recurring device fees dominate economic feasibility; physician reading burden can rapidly accumulate, but may be eased with time-based review thresholds. Specifically, we found that at clinically reported injection frequencies, hOCT was more expensive than standard monitoring for all anti-VEGF agents, with cost differences accumulating linearly over time. The expense difference at 1 year was greater with cheaper drugs (eg, bevacizumab, $9200) than with costlier drugs (eg, aflibercept 8 mg, $1892). At a range of 2 to 12 injections per year (injections cannot be given more often than every 4 weeks) with each monitoring modality, hOCT was not cost-effective for bevacizumab or ranibizumab at any injection rate combination. For aflibercept 2 mg, cost neutrality required patients receiving at least 8 injections yearly to reduce by 6 injections with hOCT; for faricimab, at least 7 injections yearly reduced by 5 injections; for aflibercept 8 mg, at least 6 injections yearly reduced by 4 injections. HOCT remained more expensive for patients receiving 7 or fewer aflibercept 2-mg injections, 6 or fewer faricimab injections, or 5 or fewer aflibercept 8-mg injections annually. Sensitivity analysis demonstrated baseline injection frequency dominated cost differences, followed by monthly device fees and achieved injection-sparing magnitude. For physician interpretation of all hOCT scans, a panel of 100 monitored patients would require 20.7 hours per week. Reviewing the single latest scan in each 30-day interval would require 0.78 hours per week (96.2% time savings); even with expedited review, a panel of 500 patients monitored with ongoing hOCT would add nearly 4 hours per week of reading burden if a physician were to review 1 scan per patient per month personally.

A central question in weighing the utility of hOCT adoption is the interpretation of value in improved quality of care and cost-saving benefits. Current literature focuses on the ability of hOCT to reduce treatment burden, but visual outcome analyses have not yet been emphasized. The independent DRCR Retina Network (formerly Diabetic Retinopathy Clinical Research Network) is conducting an ongoing phase 3 study, Protocol AO (NCT05904028), which investigates whether hOCT use results in superior visual acuity outcomes.¹⁶ If outcomes studies yield promising results, hOCT may not need to be strictly cost-effective if it improves the quality of care and outcomes available to patients. Nonetheless, an ideal implementation of hOCT would be in well-defined patient populations that support its cost-saving potential while maximizing its clinical benefit. Regardless of outcome end points, the granular information provided by hOCT will likely be clinically beneficial; a remaining challenge will be whether hOCT reimbursement is supported in a landscape of rising health care costs and increasing cost shifting to patients, especially if patients with hOCT had already been stable or maintaining good vision with existing treat-and-extend regimens.

hOCT is unlikely to fully supplant the merits of office monitoring. Compared with the focused 3 × 3-mm area captured on hOCT, standard in-office OCT devices capture at least a 6 × 6-mm view. In the office, one can also perform fundus photography, clinical examinations, and OCT- and dye-based angiography. This provides essential information on pathology in the peripheral macula and increases sensitivity to subtle levels of disease that are suitable for early intervention. Furthermore, disease progression alerts from hOCT impose new potential disturbances to clinical workflows. These events must be treated expeditiously; there may be very short notice, both for patients who are summoned to the office and for clinical staff continually shifting patient schedules, ordering drugs, and obtaining insurance approvals to accommodate a new baseline volume of unpredictable visits. Compared with the predictable rhythm of the treat-and-extend approach, physicians may need to adapt their clinics extensively to prevent the accumulation of inefficiencies.

As AI technologies scale across health care, reimbursement and litigation policy must evolve to account for the labor of managing AI data streams and absorbing liability. Current fee-for-service structures reimburse discrete interpretations but may inadequately account for time spent on algorithm supervision. Review thresholds and AI-assisted prioritization of clinically significant hOCT findings are methods to contain physician burden. However, these must still be accompanied by payment models that fairly compensate clinicians for supervising AI monitoring. Additionally, liability arises in incorporating hOCT into clinical workflows. If a physician receives a disease progression alert but is unable to schedule the patient early enough, and vision is lost, is the physician liable? Will blame for any negative outcome, negligent or not, be more likely placed on the physician if treatment frequency has been reduced and AI is performing most of the monitoring? As tort precedents and AI malpractice case law continue to develop, uncertainty may temper physician adoption and slow the integration of AI technologies into routine clinical practice.

Home OCT implementation reflects broader challenges facing AI health care technologies. Realizing the promise of these innovations requires not only technical validation but also comprehensive attention to sustainability in workflow integration, reimbursement, and medicolegal accountability in the evolving landscape of AI-augmented clinical practice. Through our research, we hope to inform and address these challenges to achieve better patient outcomes and a more sustainable model of care.

References

1. Rein DB, Wittenborn JS, Burke-Conte Z, et al. Prevalence of age-related macular degeneration in the US in 2019. JAMA Ophthalmol. 2022;140(12):1202-1208. doi:10.1001/jamaophthalmol.2022.4401
2. Desai S, Sekimitsu S, Rossin EJ, Zebardast N. Trends in anti-vascular endothelial growth factor original Medicare Part B claims in the United States, 2014-2019. Ophthalmic Epidemiol. 2024;31(5):468-477. doi:10.1080/09286586.2024.2310854
3. Heier JS, Khanani AM, Quezada Ruiz C, et al; TENAYA and LUCERNE Investigators. Efficacy, durability, and safety of intravitreal faricimab up to every 16 weeks for neovascular age-related macular degeneration (TENAYA and LUCERNE): two randomised, double-masked, phase 3, non-inferiority trials. Lancet. 2022;399(10326):729-740. doi:10.1016/S0140-6736(22)00010-1
4. Lanzetta P, Korobelnik JF, Heier JS, et al; PULSAR Investigators. Intravitreal aflibercept 8 mg in neovascular age-related macular degeneration (PULSAR): 48-week results from a randomised, double-masked, non-inferiority, phase 3 trial. Lancet. 2024;403(10432):1141-1152. doi:10.1016/S0140-6736(24)00063-1
5. Korobelnik JF, Lanzetta P, Leal S, et al; PULSAR Investigators. Intravitreal aflibercept 8 mg in neovascular age-related macular degeneration: ninety-six-week results from the randomized phase 3 PULSAR trial. Ophthalmology. 2026;133(1):39-50. doi:10.1016/j.ophtha.2025.08.022
6. Malhotra K, Colcombe J, Patil S, Vail D, Parikh R. U.S. trends of anti-vascular endothelial growth factor use from 2017-2023: an analysis of Medicare, Medicaid, and commercial insurance. PLoS One. 2026;21(1):e0335390. doi:10.1371/journal.pone.0335390
7. Rosenberg D, Deonarain DM, Gould J, et al. Efficacy, safety, and treatment burden of treat-and-extend versus alternative anti-VEGF regimens for nAMD: a systematic review and meta-analysis. Eye (Lond). 2023;37(1):6-16. doi:10.1038/s41433-022-02020-7
8. Heier JS, Holekamp NM, Busquets MA, et al. Pivotal trial validating usability and visualization performance of home OCT in neovascular age-related macular degeneration: report 1. Ophthalmol Sci. 2025;5(5):100772. doi:10.1016/j.xops.2025.100772
9. Schneider EW, Heier JS, Holekamp NM, et al. Pivotal trial toward effectiveness of self-administered OCT in neovascular age-related macular degeneration: report 2-artificial intelligence analytics. Ophthalmol Sci. 2024;5(2):100662. doi:10.1016/j.xops.2024.100662
10. Leng T, Leung EH, Mukkamala SK, et al. Longitudinal validation of the artificial intelligence algorithm in home OCT for age-related macular degeneration-report 3. Ophthalmol Sci. 2025;6(2):100907. doi:10.1016/j.xops.2025.100907
11. 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. doi:10.1016/j.ophtha.2016.04.005
12. Kim JE, Tomkins-Netzer O, Elman MJ, et al. Evaluation of a self-imaging SD-OCT system designed for remote home monitoring. BMC Ophthalmol. 2022;22(1):261. doi:10.1186/s12886-022-02458-z
13. Blinder KJ, Calhoun C, Maguire MG, et al; DRCR Retina Network. Home OCT imaging for newly diagnosed neovascular age-related macular degeneration: a feasibility study. Ophthalmol Retina. 2024;8(4):376-387. doi:10.1016/j.oret.2023.10.012
14. 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. doi:10.1016/j.oret.2022.02.011
15. Heier JS, Liu Y, Holekamp NM, et al. Clinical use of home OCT data to manage neovascular age-related macular degeneration. J Vitreoretin Dis. 2024;9(2):158-165. doi:10.1177/24741264241302858
16. Home OCT-guided treatment versus treat and extend for the management of neovascular AMD (AO). ClinicalTrials.gov. Updated January 29, 2026. Accessed March 27, 2026. https://clinicaltrials.gov/study/NCT05904028