Scientists have developed a video game aimed at diagnosing dementia before it would otherwise be detected.

The technology also tracks patients’ progress and can alert doctors if it notices any serious decline – meaning patients could get help when they need it instead of having to wait months between appointments.

Researchers at King’s College London are testing the ability of the software – called Thymia – to detect conditions like dementia, Alzheimer’s disease, Parkinson’s disease and Lewy body dementia. They plan to start clinical trials in NHS patients this summer.

Neuroscientist Dr Emilia Molimpakis, the co-founder and chief executive of Thymia, said: “In the early stages of a disease, clinicians can struggle to differentiate [these conditions] from depression but this technology can clearly distinguish between types.

“It means we will be able to detect early signs of these conditions and monitor how they deteriorate over time.”

‘Micro-expressions’

Patients can use the programme at home on any smart device or computer with a webcam and internet connection.

It asks them to play simple games, like tapping an animated bee character on a screen or verbally describing a picture, while monitoring their voice, eye movements and “micro-expressions” – tiny movements of the face.

Combining behavioural data about how well someone can complete a task with how they speak and move gives the artificial intelligence technology a detailed profile of their cognitive ability – known as their personal “signature”.

If a patient’s profile is very similar to the typical “signature” seen in people with an early form of condition like dementia, it is likely they have the disease.

Someone with a condition like mild cognitive impairment – often a precursor to dementia – may be less likely to be able to divide their attention between moving objects on a screen than a healthy person of the same age, for example.

They might also speak in slightly disjointed sentences and have less ability to maintain continuous eye contact or have microscopic facial twitches.

While none of these symptoms might be obvious to a human observer, the technology can pick up tiny differences in ability between the patient and a healthy person – and it knows that all these features in combination are a tell-tale sign something is wrong.

“Combining data from three streams – speech, movement and behaviour – gives us a much clearer picture than looking at these in isolation,” Dr Molimpakis said.

“When you combine all of these aspects, we get very strong signatures for conditions like Alzheimer’s and Parkinson’s disease.”