Research involving the University of Strathclyde examined the effectiveness of the technology in identifying people who have a genetic tendency towards the disease, years before any symptoms show.
A system called ViewMind Atlas, which uses eye-tracking and software to provide functional analysis of brain health, was examined by the study.
The system was used to detect carriers of the mutation and was 100% accurate for those who were already displaying symptoms, with 96% accuracy for those who were asymptomatic.
Participants in the research were drawn from extended families in Colombia.
An eye-tracking AI model was used to help distinguish between groups of people according to the presence or possibility of Alzheimer's disease.
The study found the tests are more accurate than traditional cognitive tests, which often do not detect Alzheimer's until its symptoms become more apparent.
Lead author Professor Mario Parra Rodriguez said: "Most of the diagnostic approaches used in dementia are expensive and invasive, because they require injection of chemical radiotracers or extraction of fluids from the body that normally require hospital settings.
"ViewMind Atlas is helping us to predict Alzheimer's dementia in people years before it becomes symptomatic.
"Doctors will receive people who are probably not yet at the stage of dementia, but they are noticing that something is going on and may have a family history.
"One of the greatest puzzles that providers face is to decide what the potential problem could be; is it age-related forgetfulness and not necessarily dementia?
"Or do these cognitive problems point towards a risk of dementia in the future, so that they could act promptly rather than waiting until the person is symptomatic?"