Fewer retests for a VR glaucoma exam

Alfaleus makes a VR version of the Humphrey visual-field test, which doctors use to check for glaucoma. The test only works if the patient keeps looking at the centre. Over 10 to 15 minutes, patients' gaze drifted and nobody noticed, so most had to take it two or three times. I added eye tracking that pauses the test when that happens.

Client: Alfaleus Technology, Hyderabad
Role: Designer and developer, contract
Timeline: Aug 2022 – Jun 2023
Product: An existing VR visual-field test, plus a VR app for low-vision patients

Role and team

My role Designer and developer, on contract.
Team [TODO: team by role]
Duration Aug 2022 – Jun 2023.
Outcome One attempt per patient instead of 2 to 3, and about 20 minutes per patient instead of 40.
Feedback Feedback that changed the design: [TODO: one real example]

2–3 → 1

Attempts per patient

Most patients needed 2 to 3 attempts before, sometimes 5

~40 → ~20

Minutes per patient

Total testing time, across every attempt

2 s

To resume

Looking back at the centre restarts a paused test

Diagram. The test runs while gaze is on the centre. When gaze leaves, the screen blanks and a pause prompt asks the patient to look at the centre. Two seconds at the centre resumes the test. Before: 2 to 3 attempts, about 40 minutes. After: one attempt, about 20 minutes per patient.

(01) The problem

A test that depends on keeping still.

In a visual-field test, the patient keeps their gaze fixed on the centre for the whole test. In Alfaleus's VR version that takes 10 to 15 minutes. Patients got restless, and their gaze drifted. Neither the system nor the doctor noticed when it happened.

So the test had to be run again. Most patients needed 2 to 3 attempts, sometimes 5. That came to about 40 minutes per patient.

Problem statement

Patients' gaze drifted off the centre during a 10 to 15 minute test, without the system or the doctor noticing, so most patients had to repeat the test two or three times.

(02) Research

Asking the doctors who ran it.

This was an existing product, so I started with the people already using it. I interviewed doctors at AIIMS New Delhi and Pondicherry Government Hospital who ran the VR test with their patients. The drifting gaze, and the retests it caused, came out of those conversations.

(03) The fix

Pause the test when the eyes wander.

I added eye tracking inside the VR headset. When the patient's gaze leaves the centre, the screen blanks and a pause prompt appears. The patient looks at the centre for 2 seconds, and the test resumes.

[TODO: image — the pause prompt inside the headset]

Fig 1. The pause prompt that appears when gaze leaves the centre.

Decision

Blank the screen, then wait for the eyes. The test stops showing anything while the patient isn't looking at the centre, and resumes on its own after 2 seconds of looking back.

(04) Outcome

One attempt per patient.

Attempt per patient

Down from 2 to 3 for most patients, sometimes 5.

~20 min

Per patient

Down from about 40 minutes.

(05) A smaller project

A VR app to help low-vision patients see.

I also designed and built a VR app for partially sighted patients, to enhance what they can see. It has:

Adjustable contrast
Zoom
Concave and convex lens simulation
Colour profiles the patient can switch between

[TODO: image — the low-vision app: contrast, zoom, lens simulation and colour profiles]

Fig 2. The low-vision app.

[TODO: images, what the pause prompt looked like, any other details]