Eye movement perimetry could speed glaucoma triage where standard testing is scarce
Glaucoma care in low-resource settings often fails at the first diagnostic step: finding patients who need formal visual field testing quickly enough to preserve vision. In Tanzania, investigators examined whether eye movement perimetry (EMP), a tablet-based test that tracks gaze instead of relying on conventional response buttons, could fill part of that gap as an initial triage tool for glaucoma.
The appeal is straightforward. Standard automated perimetry remains the reference method for assessing visual fields, but it is slow, operator-dependent, and difficult to deploy outside major eye centers. In much of sub-Saharan Africa, where glaucoma blindness is more common and disease may progress faster, those barriers can leave patients undiagnosed until damage is advanced and irreversible.
EMP aims to simplify the process by using eye-tracking technology and a portable device to estimate whether a patient has a significant field defect. In practice, that means a potentially cheaper and more transportable test, one that could be used to separate patients who likely need full perimetry from those who may not.
High sensitivity, but a narrow path to clinical usefulness
The study evaluated EMP in a real-world setting and compared its performance with standard automated perimetry across 370 eyes for correlation analysis. EMP hit rate correlated with SAP mean deviation, with Spearman’s rank correlation coefficient of 0.441 and a p value below 0.001. In imaging and perimetry terms, that is not a perfect association, but it is enough to suggest that the eye movement signal is capturing meaningful information about visual field function rather than random noise.
One sentence captures the clinical tension: EMP identified severe loss very well, but it also sent many patients onward for confirmatory testing.
In emulated triage among 232 participants, EMP achieved 100.0% sensitivity for identifying eyes with mean deviation at or below −6 dB, with a 95% confidence interval of 95.5% to 100.0%. In other words, a high EMP hit rate strongly argued against severe field loss. For triage, that is the central value proposition: a screening tool should rarely miss the people most at risk of advanced glaucoma.
Specificity was a different story. EMP specificity was 20.7%, with a 95% confidence interval of 15.0% to 27.8%. That means many individuals without a severe visual field defect would still be classified in a way that triggered referral or confirmatory testing. In a health system already constrained by limited specialist capacity, the false-positive burden could quickly blunt the method’s utility unless the workflow is refined.
Test failures and processing problems cut into performance
The practical limitations were not only diagnostic. EMP test failure occurred in 75 of 480 eyes, or 15.6%, and export or processing failure occurred in 35 eyes, or 7.3%. Those numbers matter because a triage test is only as useful as its ability to return a result consistently in the field, under routine clinical conditions, with the kinds of patients and equipment that real clinics actually have.
In this setting, failures weaken the path from a promising prototype to a serviceable screening system. Even a highly sensitive tool can create operational friction if it frequently fails to generate interpretable output, especially when the entire point is to reduce dependence on scarce equipment and specialist time.
The localization data were more encouraging. Graphical comparisons showed good matching between EMP findings and the location of visual field defects on standard testing, suggesting the device is not merely detecting that something is wrong but is also capturing where the abnormality lies. That matters mechanistically: if the test is reading the patient’s gaze shifts accurately, it may be able to reflect scotomas through altered fixation behavior, giving clinicians a biologically plausible signal rather than a crude proxy.
Patients preferred the tablet-based test over standard perimetry
Participant experience favored EMP. Patients preferred it to SAP and agreed that it was easy to perform. That preference is not a minor footnote in glaucoma care, where repeated testing is often necessary and adherence suffers when the examination is long, uncomfortable, or confusing.
A shorter, more tolerable test can change the way screening feels at the bedside. In busy clinics, especially in lower-resource environments, acceptability can determine whether a test is used at all. A method that is technically acceptable but dreaded by patients may never reach the people who need it most.
The investigators framed EMP as a potentially useful first-pass tool for identifying patients without a significant visual field defect. The logic is sound: a triage test with very high sensitivity can reduce missed severe disease, while those who screen positive can be sent to standard automated perimetry for confirmation. The challenge is the tradeoff between safety and efficiency. With specificity as low as it was here, the system would still generate a substantial number of follow-up referrals.
Where EMP fits, and what still has to change
Glaucoma detection in Tanzania is hampered by a familiar set of structural barriers: cost, portability, operator training, and the need for repeated tests. EMP is attractive because it addresses several of those barriers at once. A portable eye-tracking platform can be moved where standard perimetry cannot, and the test can be completed more quickly than SAP because it uses fewer stimuli.
Even so, the present performance profile leaves unresolved questions about deployment. The data support EMP as a candidate triage tool, not a replacement for standard perimetry. Any clinical use would have to account for its high false-positive rate and the proportion of failed or unprocessed tests. In settings with very limited referral capacity, those shortcomings could generate extra work rather than relieve it.
The study points to a realistic development path: refine the algorithm, reduce technical failure, and validate whether specificity can improve without sacrificing the excellent sensitivity seen here. If that balance can be achieved, EMP could become a practical entry point into glaucoma care for patients who would otherwise remain undiagnosed until the optic nerve damage is far advanced.
For now, the message is more constrained. EMP can rapidly identify many patients who are unlikely to have severe visual field loss, but it still sends too many others into the diagnostic pipeline. In glaucoma triage, that is the difference between an interesting device and a clinic-ready tool.