The promise of big data and artificial intelligence in healthcare has been vast for some time now, but the promise has largely remained just that as pilot projects fail to scale up.  A recent partnership between The Alan Turing Institute and UCL Hospitals NHS Foundation Trust (UCLH) is hoping to improve matters.

‘The NHS routinely collects data that is analysed to develop research, track performance and measure outcomes but we could do so much more with the information we collect. Imagine a world where we could use this data to develop algorithms to rule out diseases, suggest treatment plans or predict behaviour….that is more than possible with the wealth of data we have available and the expertise at The Alan Turing Institute. The partnership has the potential to tackle some of the big issues that the NHS has never been able to solve,” UCLH say.

Improving A&E

The first area of work to be tackled by the partnership is the accident and emergency ward.  A&E waiting times regularly exceed four hours across the UK, despite the best efforts of staff.  The partnership will utilize the data science capabilities of the Turing Institute to try and improve the service for patients and staff.

“Imagine a scenario where patients present to A&E with abdomen pain – our standard response is to check bloods, order X-rays or scans and in probably about 80% of cases, discharge for home management.  What, if through analysis of thousands of similar scenarios, we were able to identify patterns in the initial presentation of the 20% with serious conditions, such as intestinal perforation or severe infections? This could enable us to fast track them through to a scan and a swift diagnosis and could support clinical decision making to manage the 80% who need no further clinical input more effectively. Machines will never replace doctors, but the use of data, expertise and technology can radically change how we manage our services – for the better,” the team say.

The partnership hope to better understand and improve the flow of both staff and patients through the hospital.  They will do this by using AI to examine large datasets on the movement of people through the various departments of the hospital to try and identify bottlenecks and other hurdles that cause downtime and delays in service provision.

Of course, this isn’t the first project of this nature to be undertaken around the world.  For instance, back in 2016 I wrote about a paper that utilized predictive analytics to minimize patient waiting times.  The authors propose a model that translates predictions on the arrival of patients into the emergency room into a better process for dealing with those numbers.

The Assistance Publique-Hôpitaux de Paris (AP-HP) hospital group attempted to put this kind of thing into practice last year.  The group have analyzed ten years worth of data on hospital admissions, flu rates, weather and so on by its Trusted Analytics Platform (TAP).  The aim is to better predict hospital admissions, and therefore better route people to the best facility.

TAP is an open source platform that uses machine learning to help analysis of data.  The AP-HP project is the first to use the platform for time series analysis, but should the project be successful, it will be rolled out across all of the 44 hospitals in the group.

The system aims to provide admissions staff with accurate predictions up to 15 days in advance, thus hopefully providing sufficient warning to allow suitable staffing and other resources to be on hand in periods of high demand.

It’s unclear whether the success the team hoped for materialized, but it’s an approach the Turing Institute and UCLH hope to succeed with.