ISSUE 002 · Oct 4, 2026 · 8 min read

What if retinal relapse signals could reshape the clinic schedule?

Turn AI fluid signals from self-imaged OCT into a retina response network: specialist review within one business day and an expedited visit pathway within 72 hours.

The issue in one sentence

Turn AI fluid signals from self-imaged OCT into a retina response network: specialist review within one business day and an expedited visit pathway within 72 hours.

Diagram 01

The response loop from scan to access

01Home OCT→

02AI fluid volume→

03Trend alert→

04Specialist review→

05Priority slot→

06Visit · injection→

07Vision · burden

ANM proposed model. It separates FDA-authorized visualization and quantification from physician treatment decisions, making alert review, scheduling, and follow-up the testable intervention.

01Clinical problem · Evidence

Retinal relapse begins between scheduled visits.

Anti-VEGF can control neovascular AMD, but office OCT samples disease activity sparsely. Late detection risks vision; frequent visits for everyone burden patients, caregivers, and retina clinics.

02Technology stack · Proposal

Join self-OCT, AI fluid trajectories, and scheduling in one loop.

Patients self-image at home and AI quantifies intraretinal and subretinal hyporeflective spaces. An alert does not prescribe treatment; it opens a physician review queue and an expedited appointment path.

03Research question · Hypothesis

The device can see—can the clinic act in time?

Can a service layer guaranteeing review within one business day and, when needed, a visit within 72 hours reduce fluid exposure and routine visits without worse vision than treat-and-extend?

04First study · Proposal

Eight clinics, 240 people, a 12-month pragmatic trial.

All treated adults with nAMD self-image. The intervention routes alerts to review within one day and, if needed, a 72-hour slot; control data remain masked while usual treat-and-extend continues. Co-primary outcomes are 5-letter noninferiority in ETDRS vision and retinal-fluid AUC, with visits, injections, staff time, and caregiver time measured.

05Defensible data · Venture assumption

The moat is what happens after the alert—not fluid volume alone.

Link patient-level fluid trajectories to alerts, human review, contact attempts, appointment capacity, injections, and vision outcomes. This dataset can tune safe thresholds and clinic capacity together.

06Venture thesis · Assumption

The first customer is the retina practice, not the device maker.

Offer device-agnostic monitoring and escalation to retina groups per monitored eye-month. Initial value is urgent-slot operations, documented review time, and missed-alert audit—not assumed visit replacement. Reimbursement or shared savings is a later hypothesis contingent on clinical and economic evidence.

Evidence map

The last mile exposed by the evidence

96.1%

At-home self-imaging success in a multicenter pivotal study

91.2%

Share of 4,435 scans eligible for AI fluid quantification in a small management study

15.3 weeks

Mean injection interval during monitoring in a 27-eye before–after study; 8.0 weeks before

603 people

Enrollment in the ongoing 104-week phase 3 comparison with treat-and-extend

Evidence

What is established—and what is not

Established: FDA granted De Novo authorization to an AI-based, patient-operated Home OCT in May 2024, and a multicenter pivotal study reported 96.1% at-home self-imaging success. In a nonrandomized before–after study of 15 people and 27 eyes, the mean injection interval increased from 8.0 to 15.3 weeks, while visual-acuity change was not significant. Still hypothetical: whether these findings generalize at scale and whether alert-to-review-to-scheduling operations create clinical and economic value. The 603-participant Protocol AO is following patients for 104 weeks to compare Home OCT guidance with treat-and-extend.

FDA De Novo decision Multicenter pivotal study Protocol AO registry

Failure conditions and safety, ethics, and regulatory risks

Clinical failure condition

Stop if 12-month vision crosses the 5-letter noninferiority margin or if fluid AUC or rescue visits increase, even when operations look easier.

Alert liability

Define weekend coverage, acknowledgement, escalation after failed contact, and a separate instruction for patients to report worsening symptoms immediately.

Access bias

If low vision, cognitive impairment, tremor, or low digital literacy excludes patients, benefits accrue to the easiest users. Test caregiver mode and device support separately.

Regulatory boundary

The current FDA authorization covers visualization and quantification, not replacement of routine care or autonomous treatment decisions. Treatment recommendations may trigger separate SaMD review.

Algorithm drift

Track performance by device and software version, low-quality scans, and false alerts for small fluid volumes; retain human access to raw B-scans.

Business failure condition

Do not scale if per-eye-month revenue cannot cover devices, monitoring, clinician review, and urgent capacity, or if net review time per clinic does not fall.

Bibliography

References

Established facts are separated from research and venture hypotheses that still require validation.

01
U.S. Food and Drug Administration. Notal Vision Home Optical Coherence Tomography (OCT) System. De Novo DEN230043.
Primary regulatory record confirming De Novo authorization and device classification for an AI-based home OCT system.DEN230043 · May 15, 2024
02
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.
A multicenter five-week at-home study reported a 96.1% self-imaging success rate.PMID 40458667
03
Schneider EW, Heier JS, Holekamp NM, et al. Pivotal Trial Toward Effectiveness of Self-administered OCT in Neovascular AMD. Report 2—Artificial Intelligence Analytics. Ophthalmol Sci. 2025;5(2):100662.
Validated AI estimation of hyporeflective-space volume in self-acquired OCT scans for regulatory review.PMID 39811265
04
Holekamp NM, de Beus AM, Clark WL, Heier JS. Prospective trial of Home OCT-guided management of treatment-experienced neovascular AMD patients. Retina. 2024;44(10):1714–1731.
A nonrandomized before–after study of 15 people and 27 eyes; longer injection intervals are hypothesis-generating because of its size and lack of a parallel control.PMID 39287534 · NCT05650047
05
DRCR Retina Network. Protocol AO: Home OCT-Guided Treatment versus Treat and Extend for the Management of Neovascular AMD.
The 603-participant, 104-week comparative trial was in follow-up as of October 2026; superiority should not be assumed before results.NCT05904028

The next issue publishes every day at 9 AM KST.All issues