How artificial intelligence is helping patients avoid dangerous side effects

Close up of a pair of hands, one holding a glass of water, and the other holding two pills. The rest of the blister pack is on the surface in front of them

A new artificial intelligence-powered tool developed by a primary care provider with the help of ARU experts is helping ensure that patients taking multiple medications do not suffer any harmful side effects.

Data Scientist Dr Amirali Shirazibeheshti was recruited to work with AT Medics to develop a new plug-in for the company’s EZ Analytics population healthcare management software.

AT Medics are London’s largest primary care provider, which offers services to over 390,000 patients across 49 locations. Dr Shirazibeheshti’s work was carried out under a Knowledge Transfer Partnership (KTP), and the new module is already in use helping GPs and patients.

The new function addresses issues around polypharmacy, when patients are taking different combinations of medicines to treat multiple conditions. The automated system developed under the KTP enables doctors to cross reference data on drugs with patient information, to pinpoint combinations which could potentially cause problems.

'Prescribing multiple medications can lead to the patient becoming ill through interactions between those medicines, rather than the condition they were being treated for in the first place,' says Dr George Wilson, Associate Professor in the School of Computing and Information Science at ARU.

'It’s a seriously important issue, particularly for patients who have several co-morbidities and can be taking a lot of different drugs.'

The KTP work focused on a particular class of medicine called anticholinergic drugs. They are commonly used to treat a diverse range of conditions, but are known to cause side effects when deployed in combination.

'Previously it was the responsibility of a GP or healthcare professional prescribing these medicines to manually check that there weren’t going to be any detrimental effects,' says Dr Wilson.

'It involved finding out information about all the different medicines and going through the patient’s record, so it was a very labour-intensive operation that could take hours, if not days.'

Using a technique known as cluster analysis, the tool scans over 300,000 patient records and cross-references them with an NHS repository of risk information about drugs.

'It looks for clusters of data that are in some way related, so you get groups of patients with a similar profile,' explains Dr Wilson.

'Once it has identified several clusters, it prioritises them by who is most at risk, so when a GP uses this technology, they are able to see how many patients could suffer side effects from a particular drug combination.'

Doctors can then use this information to make choices about which medicines to prescribe.

'Within a few seconds you have a result without having to trawl through hundreds of patient records,' says Dr Wilson.

Knowledge Transfer Partnerships allow ARU to work with businesses in order to solve specific problems. The process involves appointing a suitably qualified graduate "associate", who is employed by ARU but based in the business to help the KTP address its challenge.

ARU had worked with AT Medics as a Knowledge Transfer Partner on previous projects, and Dr Wilson says the partnership was beneficial for both parties.

'The company gets a software product, and as a university we get research outcomes,' he says. 'Last year we got a conference paper published on this work, and we’ve also submitted a paper to a medical journal.

'The success of this project can help us obtain further research funding, which is good for us as academics and good for the University.'

Dr Wilson, who led the project with his academic colleague Dr Cristina Luca, paid tribute to the work of Dr Shirazibeheshti.

'Amirali had to learn a lot about cluster analysis and about Python programming which is a commonly-used scripting language for accessing and processing data,' he says.

'The mentors he worked with at AT Medics were extremely supportive of his work, and it’s a great credit to him that he was able to successfully walk the line between being an academic who has to be research-orientated, whilst also meeting the needs of a company which is looking for a specific result.'

The new software module is already in use as part of the EZ Analytics platform, which is deployed at AT Medics clinics and licensed to other healthcare providers via the company’s medtech arm, AT Tech.

'When we put the tool out to GPs to try the new functionality, the feedback was tremendous,' says Dr Wilson. 'It’s already having a big impact for doctors and their patients.'

Learn more about how a Knowledge Transfer Partnership can help your business.