Around 850,000 people are currently living with dementia in the UK and this number is predicted to rise with increasing life expectancy in future. By some calculations, one in three people born in the UK in the last few years will go on to develop dementia in their lifetime.

Yet despite this, the condition is poorly diagnosed. "We don't have an objective, sensitive tool that can pick it up at the early stages," says Sina Habibi, founder of a Cognetivity, a health tech startup looking to address this problem with an entirely new solution.


In the US, 50 percent of those with the degenerative condition are not officially diagnosed, and this figure is about 30 percent for the UK. And those that receive a diagnosis generally do so only when they are already severely debilitated and facing a precipitous decline. This ramps up the cost of treatment as well as dashing hopes of decelerating onset with a programme of care. 

One reason for this, according to Habibi, is the method of diagnosis used.

"The trigger for assessing someone for Alzheimer's and dementia is GP suspicion, but there is no definition of GP suspicion," says Habibi of the NICE clinical guidelines used in UK medical practice. "You can see that this becomes very subjective, very crude," he continues, adding that diagnoses are normally reliant on self reporting where people ask their GP to be referred to secondary care.

The difficulty, according to Habibi, stems from the fact that most tests are based on assessing memory, as this is one of the earliest and most obvious faculties to go. But he believes this is the wrong approach: "Memory is difficult, it's subjective, it takes a long time to assess."

Instead, he and his cofounder Seyed-Mahdi Khaligh-Razavi have come up with a new measure: image cognition.

Khaligh-Razavi is a computational neuroscientist, and his PhD focused on how human vision worked by placing participants inside MRI and fMRI scanners. Part of his research examined how quickly it took participants to descramble a distorted image, of the type you may see as a captcha on a website. He noticed a pattern: the speed with which people could do so decreased with age. 

"In other words, how quickly we interpret images and understand the information in visual format," says Habibi. 

"We knew we'd found something very interesting. We basically started with the solution and then looked for a problem big enough to solve it with."

At this pivotal point Habibi, mechanical engineer by background, was enrolled in a PhD in nano biotechnology.

"I realised I wasn't happy with the pace of research in academia and I wanted to apply my skill set to the world of ventures which is faster, more eventful and exciting," says Habibi. He became chair of the Cambridge University Entrepreneurs society, where he met Khaligh-Razavi.

"We wrote back-of-the-envelope code together, put it to test, realised it worked and applied for a patent," he explains. They founded the company almost exactly six years ago, in April 2013.

While the initial research focused on distorted images, at present Cognetivity's test involves presenting non-distorted images of the natural world to participants and asking them to indicate which category they belong to. Artificial intelligence algorithms are then used to cluster the results by speed, accuracy and image type and signal whether a participant is in the danger zone. 

But why is a test of image cognition particularly effective at flagging up neural degeneration of the type exhibited in dementia? "There are a few different theories," says Habibi. "One is that the visual cortex, which is in charge of breaking down information in the form of images, is the area of the brain where deterioration starts first."

However, Habibi says that Cognetivity's test is not limited to the visual cortex, as respondents are also required to touch a screen to indicate a response, the motor cortex is also stimulated.

"You've got the hippocampus involved, you have frontal lobe involved - most of the brain is activated and that makes it very difficult for the brain to mask its problems," he says.

At present, the technology is delivered by Apple's iPad "because the timing is very, very important in this test," says Habibi. "We're working on milliseconds accuracies which cannot be delivered on most other tablets."

Five trials of the technology have been completed, and there is currently another trial underway at London's Moseley Hospital's lab by an independent organisation in order to achieve the desired standard of scientific verification. Habibi says the results will be communicated with the requisite regulatory bodies - the MHRA in the UK and the FDA in the US. At that point, they will be able to release the product to the market.

He says the team is collaborating with four NHS trusts and that they will "naturally be the first customers". Despite the fact that the tool can be operated on an iPad, Habibi says it will still be necessary for patients to complete the test in a GP's surgery. "We cannot put diagnosis to the hands of patients, this is something that we cannot do by law," he explains.

However, Habibi says there will be scope for patients to monitor the progression of their condition within the comfort of their homes: "It will be better for GPs and patients, and a huge cost-saving exercise."