Artificial intelligence and telemedicine change how patients move through eye care pathways. In ophthalmology, digital fundus photography supports screening for diabetic retinopathy (DR), glaucoma, age-related macular degeneration (AMD) and cardiovascular risk assessment from retinal images. A Journal of Medical Internet Research viewpoint compares these developments in the United Kingdom and Germany. AI has entered ophthalmic screening since around 2018, with regulatory approvals covering DR, glaucoma, AMD and emerging oculomics tasks. Its effect depends on pathway design, not only diagnostic performance.
AI Needs Structured Pathway Design
Ophthalmic AI can sit in several positions within a patient journey. Conventional care keeps the first clinical assessment with an ophthalmologist, while AI may support the physician in the background. A tele-ophthalmology or grader-based model uses image grading or remote consultation before deciding whether specialist referral is needed. A semiautomated model lets AI provide the initial assessment, followed by human review before in-person care. A fully autonomous model allows AI to make the primary screening decision and determine which cases require physician assessment.
These distinctions matter because screening, surveillance, triage, diagnosis, referral, treatment and follow-up are separate parts of care. AI may support autonomous diagnostic tasks, remote image grading or surveillance over time, but its role must match the intended downstream action. Digital-first routes can improve access when nonmedical staff capture images in primary care, community clinics or retail settings. They can also create diagnostic dead ends when abnormal, uncertain or low-quality results lack a clear escalation route.
Structured information transfer is central to safe implementation. AI reports need more than a binary referral flag. Useful outputs include image quality, suspected diagnosis and severity, urgency, recommended follow-up interval and relevant systemic risk factors.
Must Read: AI Use in Hospital Medicine Needs Better Implementation
UK Pathways Build on Gatekeeping
The United Kingdom’s National Health Service (NHS) provides universal coverage and generally free care at the point of use. Access to hospital ophthalmology is controlled through a gatekeeper model, with primary care providers rather than patients determining specialist referral, except for acute emergencies. Community optometrists therefore play a major triage role. Patients with visual symptoms or those attending routine refraction usually see a community optometrist first, often in a retail setting, before referral to hospital eye services if pathology is suspected.
DR follows a more organised pathway through the NHS diabetic eye screening programme (DESP). Eligible people with diabetes attend community clinics or mobile units for regular screening. Technicians capture digital fundus images, which are transmitted to a central grading centre and reviewed by trained human graders under a national protocol. England is validating autonomous grader-based filtering models in which AI excludes images without disease, while Scotland has used an automated first-line grading system since 2011.
For neovascular AMD, the pathway remains symptom driven and time critical. Patients commonly present to optometrists with distortion or central vision loss and fast-track electronic referral aims for specialist assessment within two weeks. Glaucoma detection remains largely opportunistic through routine sight tests, with some regions using specialist community optometrists to refine referrals before hospital escalation. Oculomics for cardiovascular risk assessment is emerging mainly as a self-pay add-on in private optometry and clinics.
Germany Shows Fragmented Adoption
Germany’s eye care system differs through direct access to office-based ophthalmologists. Statutory health insurance provides near-universal coverage, while outpatient eye care is delivered by office-based ophthalmologists reimbursed by sickness funds. Patients can consult an ophthalmologist directly. Opticians may perform screening tests such as tonometry, perimetry and anterior or posterior segment examinations, but suspected abnormalities require referral to a physician for formal diagnosis. The standard pathway remains physician based, with the final screening decision, diagnosis and management performed face to face by an ophthalmologist.
DR screening is opportunistic rather than organised through a national tele-ophthalmology programme. National guidelines mandate annual or biannual eye examinations for patients with diabetes, but responsibility lies with patients and diabetologists to ensure visits occur. Screening is performed by office-based ophthalmologists using slit lamp biomicroscopy with or without fundus photography, while telemedicine and AI integration remain limited to pilot projects. Retail-based screening has introduced an additional semiautomated, grader-based route.
For neovascular AMD, patients with visual changes typically self-refer to local ophthalmologists. Many office-based ophthalmologists provide intravitreal injections or work with specialised injection centres. AI tools may assist with AMD biomarker segmentation and classification, but pilot use does not alter the patient journey. Glaucoma detection is often commercially driven because statutory insurance does not routinely cover asymptomatic screening. Retinal cardiovascular risk assessment is confined to exploratory retail offerings, without formal integration into funded pathways or defined referral arrangements.
AI and telemedicine can create markedly different ophthalmology journeys in the United Kingdom and Germany. The same technologies may support organised screening, community triage, physician decision support or retail-based self-pay services depending on health system structure. Safe scaling depends on pathway design that defines where AI sits, who receives its outputs, how follow-up is organised, how reimbursement works and how responsibility is shared. Clear escalation routes, structured records, processes for ungradable images and appropriate task shifting are needed to improve access without weakening safety or accountability.
Source: Journal of Medical Internet Research
Image Credit: iStock
References:
Poschkamp B, Ong AY, Engelmann J et al. (2026) Rerouting Eye Care: How AI and Telemedicine Are Reshaping Ophthalmology Patient Journeys in the United Kingdom and Germany. J Med Internet Res;28:e93170.
