← All work
Academic project · IIT Gandhinagar · Summer 2018

Before you can screen with gaze, you have to understand the paper tests.

A summer research internship in a cognitive science lab building a gaze-based tool to screen for dementia. My job was the groundwork: running the standard paper-based screening tests on healthy participants, so the team had a baseline to compare against.

My roleSummer research intern, M.Sc. Cognitive Science
TimelineSummer 2018 · 9 weeks
LabIRACS Lab, IIT Gandhinagar
MethodsMoCA, MMSE, ACE-III, participant recruitment
60
Participants in a pilot study, recruited and tested by me
3
Standard dementia screening scales administered and scored
~40 min
Per participant to run all three, before feedback
01

The projectA lab building a faster way to screen

Dementia is usually screened with pen-and-paper tests that need a trained person, a quiet room and a lot of time. The lab I joined was developing a gaze-based screening system that uses a camera and image processing to read eye movements instead.

To know whether gaze could stand in for the paper tests, the team first needed solid data from those tests. That was my part of the project.

The question

How do the standard dementia tests work in practice, and how do results differ between controlled lab settings and real patients?

02

What I didRan the paper tests, end to end

1Learned the scalesTrained by a senior researcher to administer and score MoCA, MMSE and ACE-III.
2Recruited participantsPersonally recruited 60 healthy student participants for the pilot over three weeks.
3Ran the sessionsAbout 40 minutes per person across all three tests, followed by informal feedback.
4Scored by domainMemory, attention, visuo-spatial, language and more. The data went to my teammates for analysis.
The lab's gaze-tracking setup: chin-rest, camera and monitor
The lab’s gaze-based setup: a chin-rest, a depth camera and a monitor. I gained hands-on experience with it alongside the paper tests.
03

What I learnedWhy paper tests were the problem worth solving

They take a long time

Around 40 minutes for three tests, and participants’ patience ran out before the end.

Scoring isn’t consistent

Different researchers scored the same tests differently, which makes results harder to compare.

The setting changes results

Noise in the lab and time pressure affected how people performed.

Each scale has a place

MMSE is quick but limited. Longer tests give more detail when there’s time. The choice depends on the setting.

What it taught me

This was my first look at how a measurement tool shapes the answer. A test is only as good as the conditions it’s run in, and that’s why a faster, more objective screen was worth building.

Cognitive assessmentParticipant recruitmentLab study protocolEye-tracking setupData collection

Academic internship project. The setup photo is cropped so no participant can be identified.