The Evidence for AI and VR in Eye Care Diagnostics

In our first blog in the AI and VR in eye care series, we explored how AI and VR can help address long patient wait times and diagnostic bottlenecks in eye care. But are these innovations truly reliable? In this piece, we look at the growing body of evidence showing how AI-driven analysis and VR-based diagnostics are enhancing early detection, improving efficiency, and reshaping patient care without replacing the clinician’s expertise.

‍

Evidence Roundup: Automated Perimetry, Virtual Reality Visual Field Tests, and Outcomes in Glaucoma Care

Over the past few years, as technology and AI have been further studied and developed, the eye care industry has grown to incorporate AI and VR technology into diagnostic equipment. AI and VR diagnostic tools can serve as powerful assistance to clinicians if adopted and appropriately regulated. The idea is that AI and VR tools enhance the care a doctor can provide, rather than replace it. In addition to the benefits an eye care practice achieves, patients receiving care experience faster, more accessible treatment. 

According to the Centers for Disease Control and Prevention, “approximately 12 million people 40 years and over in the United States have vision impairment.” Additionally, more individuals are living with chronic health conditions that contribute to vision loss. As demand for eye care increases, advanced technologies such as AI and VR can provide timely, convenient diagnostics. Most eye diseases are diagnosed through scans, images, or magnetic resonance imaging (MRI). AI can be trained to recognize early anomalies of these diseases and help clinicians detect them more quickly. 

In April 2018, the FDA approved IDx-DR, the first AI device capable of screening for diabetic retinopathy. IDx-DR is software that interprets and analyzes uploaded scans and images to identify signs of diabetic retinopathy. 

According to the American Academy of Ophthalmology (AAO), “AI systems are capable of detecting diabetic retinopathy with an accuracy that rivals or even surpasses that of experienced ophthalmologists.” Furthermore, the EyeArt AI system demonstrated Sensitivity: 96%, Specificity: 88% in diagnosing diabetic retinopathy across more than 850,000 fundus images, rivaling expert-level performance.

By incorporating advanced automated technology into the early assessment of a patient, doctors can gain a general understanding of the patient's condition before their appointment. 

Previously, the retinal exam required a visit to an eye specialist; however, technology like IDx-DR now saves time and helps determine when to refer a patient to a specialist. With diagnostic equipment that detects anomalies, doctors gain an advantage and can begin developing a treatment plan. Even outside of eye care, AI-assisted diagnostics have reduced outpatient wait times from a median of 1.97 hours to just 0.38 hours, demonstrating the broader potential of AI integration to streamline workflows.

Additionally, the use of virtual reality in diagnostic testing is transforming the field. Providing diagnostic tests that are both physically and financially accessible while engaging the patient creates a win-win situation. Although AI-based diagnostics might have higher upfront costs — for example, $559 per patient compared to $533 with traditional screening methods — they provide long-term savings through better access, earlier detection, and faster results. 

Cost-effectiveness should be viewed holistically, considering both financial efficiency and health outcomes. Doctors can reach patients who previously could not take standard tests, and a head-mounted perimeter with functionally indistinguishable results to the HFA offers a low-cost, portable alternative to specialized clinic-based perimetry devices.

‍

Addressing Concerns: Reliability, Transparency, Ethics, and Regulatory Considerations

Despite notable progress, doubt about AI and virtual reality persists. A prevalent concern is the perceived “black box” characteristic of AI tools. While algorithms produce results, not every clinician is confident in grasping how results are formulated. As the American Medical Associates notes, "growth and adoption could stall if physicians aren’t told what the technology is doing and how it’s doing it or if they are unable to explain the functions to their patients," underscoring the need for transparency in AI in medical settings.

There is also concern regarding AI’s diagnostic accuracy and whether it can truly match human expertise. While studies show that AI can detect diseases like diabetic retinopathy, glaucoma, and macular degeneration with impressive sensitivity, some clinicians worry about subtle anomalies being overlooked, risking an early diagnosis. However, comparative studies show that AI performs on par with, or better than, clinicians in estimating diagnostic probabilities. One JAMA Network study found AI systems produced lower error rates than clinicians across multiple conditions.

In the case of VR-based diagnostics, the concern lies in whether virtual methods truly replicate the standards of traditional devices, such as the Humphrey Field Analyzer. 

A study was conducted to compare Virtual Field’s VR headset with the Humphrey Field Analyzer. The conclusion was, “Virtual visual field testing using the BOLT strategy was similar to the Zeiss Humphrey SITA-Standard 24-2. VVF testing provides advantages over HVF in terms of cost-effectiveness, portability, and efficiency.” 

‍

As AI and VR technologies take on more responsibilities in eye care, regulatory oversight is both necessary and foundational. AI and VR tools should comply with healthcare standards such as the Health Insurance Portability and Accountability Act (HIPAA) to protect patient data privacy. In the United States, devices and software used for diagnosing or influencing clinical decision-making are subject to FDA approval, ensuring their safety, efficacy, and transparency. However, since AI algorithms are constantly learning and evolving, this must not be a one-time checkpoint.

Additionally, there are discussions about the need for transparency in explaining AI diagnostics, ensuring that clinicians and patients understand how the diagnostic result was generated. Without transparency, trust in AI systems remains fragile, making the role of regulation not just protective but restorative. 

While these concerns are valid, as technology evolves, we must address skepticism through education, transparency, and inclusion in the development and implementation processes. Rather than replacing clinicians, AI and VR should be viewed as tools that enhance clinical capacity, reduce administrative burdens, and support earlier and more accurate diagnoses, particularly in underserved areas.

‍

How Do We Put This Evidence Into Action?

The growing body of research makes one thing clear: AI and VR are no longer experimental concepts. Rather they are proven, scalable tools that can strengthen diagnostic accuracy, reduce wait times, and expand access to care. But evidence alone isn’t enough. The real challenge lies in turning these capabilities into everyday practice through thoughtful planning, clinician engagement, and strong data governance. So how do eye care organizations move from possibility to practical implementation? 

In the final post in this series, we’ll provide a step-by-step roadmap for safely and successfully integrating AI and VR into clinical workflows, from readiness assessments and pilot programs to full-scale rollout and long-term sustainability.