Dementia diagnosis could be made earlier if use of robots in GP surgeries becomes the norm

But while family doctors are unlikely to piece together the scattered clues and spot the warning signs of dementia early, artificial intelligence is much more likely to make the connections.

A team working at Plymouth and Edinburgh Universities found AI had an 84 per cent success rate in spotting early warning signs of dementia.

The research holds out the possibility that dementia cases could be diagnosed as much as a decade earlier than at present.

And that means tens of thousands of patients could get earlier treatment aimed at slowing down the progression of the condition.

Dementia – a loss of memory and thinking skills that eventually renders the patient unable to perform everyday tasks – currently affects 850,000 people in the UK.

In many cases, the disease will have been progressing for 10 – 15 years before it becomes severe enough to be spotted by a GP and diagnosed by a specialist.

The study collected GPs’ clinical and administrative data from 18 consenting practices across Devon for 26,483 patients over the age of 65.

Dr Javier Escudero, Chancellor’s Fellow in Biomedical Signal Processing at the University of Edinburgh, said: “The idea of this feasibility study was to check whether a machine-learning based system could use data routinely available at GP practices to identify people who would be at a higher risk of dementia.”

Robots could be used to help treat patients (Image: GETTY [stock image])

Dr Escudero explained that the team created a profile of conditions, diseases and behaviours associated with dementia, which also include “high fragility” incidents such as falls and forgetfulness.

He said: “We then use machine learning techniques to try to identify people who appeared to have that profile of dementia but had not received a formal diagnosis.”

The AI scanned each patients’ notes and was “triggered” by one or two potential indicators.

The flagged cases were then checked in more detail by a GP, with 84 per cent of the checks confirming previously undiagnosed cases of dementia.

The technology is something that could be run in a GP practice and used to ring alarms, saying that this person could be at risk of dementia.

Dr Javier Escudero

Dr Escudero, said: “The diagnosis right now depends on the person showing symptoms of the disease by that time the brain is already damaged.

“By diagnosing dementia early it allows the people to, if they like, prepare and access special care they may need to help improve the quality of their life.

“It allows the GP or NHS to phone up the patient to explain the benefit of special care as the treatment can help the patient manage the symptoms and experience a better quality of life.”

He continued: “The technology is something that could be run in a GP practice and used to ring alarms, saying that this person could be at risk of dementia.

“We don’t want to replace GPs with machines we wanted to help them by creating something that could alert them to the disease.

“It still needs a proper clinician’s judgement and it is still very important to identify these people with dementia.”

He added: “We now know that neurodegeneration takes 10-15 years. If machine learning and other data are used it will be able to detect dementia three to five years into the start of the brain’s decay.”

Professor Helen Stokes-Lampard, Chair of the Royal College of GPs, said: “AI and other technological advances have huge potential to support general practice, and indeed healthcare as a whole, as long as they are implemented equitably and in the best interests of patients, after research has proven their worth and safety, so it’s good to see high-quality research being conducted in this area in primary care.”

The research holds out the possibility that dementia cases could be diagnosed even a decade earlier (Image: GETTY [stock image])

She added: “Dementia can cause untold misery to patients, their carers and their families – and early diagnosis can certainly help in terms of taking steps to delay escalation of the disease.”

Prof Stokes-Lampard warned: “Any increase in dementia diagnosis must also be matched with adequate post-diagnostic services to refer patients to in the community, so that patients can access the most appropriate treatment – and access in some areas is insufficient as it stands.”

Jim Pearson, Director of Policy and Research from Alzheimer Scotland, said: “Alzheimer Scotland welcome this interesting research about the role that artificial intelligence may play to help GPs to diagnose dementia.

“Whilst this study is promising to support a formal diagnosis within a routine primary care setting, overall we need much more research into the causes of dementia, treatments and supports that allow people to live well with dementia, as well as seeking to prevent and even cure of the condition.”