Two NHS organisations are showing how to strengthen their capabilities with new approaches to automation, writes Philip Sheen, head of public sector at UiPath
The UK public sector is faced with a set of systemic challenges, and to overcome them it needs to find a new formula in the relationship between its people and technology – with automation providing a key factor.
Everybody is aware of the core problems: increasing demand from more people living longer, rising expectations of the quality and availability of services, and the long term squeeze on public funds.
Many of its people are doing great work, but the sector is not always getting the full value, as they spend so much time on mundane tasks and less on challenges that need their emotional intelligence and ability to take subjective action.
It needs to look closely at how it carries out its work, with a focus on how its people can provide their full value and how its technology stack can make this possible. This is where automation is crucial in supporting the change, especially as it goes beyond the already common robotic processes, adding the capabilities of machine learning and artificial intelligence.
While for many organisations this involves exploring new ground, there are some taking the automation of administrative tasks to new levels and providing lessons from which others can learn. Two NHS bodies outlined to the recent UKAuthority Bots4Good conference how they have been working with UiPath to deploy its technology to augment the human workforce.
The Plymouth approach
A systematic approach has been developed by University Hospitals Plymouth (UHP) NHS Trust, whose lead automation developer, Chris Butson, described how it has made quick progress after its first deployment of robotic process automation in 2021.
In developing a business case it quickly replaced the term ‘robotic’, which made some people feel uncomfortable, with ‘digital assistants’, emphasising that the intention was to support, not replace staff. Then after some research it identified UiPath as the platform provider with Mastek as reseller.
It began to build an in-house development team to handle coding and object oriented programming, with one lead developer and business analyst supported by two developers with a coding background. Butson said the priority was to find people who could think logically and were keen to learn. While the coders had no experience of UiPath technology, they were soon able to acquire the relevant skills through its academy.
“We started looking for some quick wins within our own department, the performance information team,” Butson said. “This allowed us to train up on using the software while using real life processes and scenarios.
“It seemed the logical thing to do rather than using dummy exercises or fake data. It allowed us not only to familiarise ourselves with the software, but also build an understanding of how a project runs, starts and finishes, including the documentation, governance, approvals, everything that’s required around an automation deployment.
“It also meant we had relationships built with the subject matter experts, so we knew what we needed from them – a clear understanding of the process – and they knew what they needed from us – an automated routine delivered.”
Small and frequent
The early emphasis was on small, frequent tasks carried out daily by the trust’s information analysts, beginning with ambulance handover and deceased reporting.
“After successfully delivering the first couple of automations, word started to spread to teams within the departments such outpatient appointments and procurement, which generated more project ideas and success,” Butson said.
The UHP communications team provided support, promoting the work of the development team throughout the trust.
Over the past two years the team has engaged with 16 others throughout the trust – in areas ranging from patient experience to payroll to cardiology – receiving 71 project submissions and taking 15 to completion. Butson said this had so far saved over 4,300 hours of manual time, releasing £81,5000 of cash and non-cash benefits, and that for these projects the figures were expected to reach 40,000 hours and £650,000 over five years.
Also important was that the automations were expected to do the work of five full time staff over the period and to pay for themselves within six months.
Projected benefits for ongoing projects take the benefits to a higher level: 33,580 hours per annum, 21 full time equivalents and £3.6 million – with the potential for other cash releasing, societal and quantitative benefits.
UHT has now built a reputation for automations that have led to talks for it to become a centre of excellence for the Devon Integrated Care Service, and to be asked by UiPath to become its NHS Autopilot site in 2024. It has also received training in the company’s Document Understanding platform and identified the potential to use this in clinical coding.
Application in finance
That technology has already been used by NHS North of England Commissioning Support Unit, whose technical architect Jonathan Boyle also presented at the conference.
He explained how it has been applied to take the manual element out of some finance processes, with the Dispatcher bot logging into the organisation’s Oracle system daily to extra details of all reports, then going through these to determine if the report should go to Document Understanding or Performer path. If the former, an invoice is downloaded, its text digitised and the desired data classified using keywords and extracted using machine learning models. If the latter, the data is extracted then added to the relevant ledger.
The process takes account of the need for a minimum confidence threshold in the extraction of at least 90%. If the robot cannot do this the document is directed to the UiPath Action Centre, where it can be examined by a human and relabeled.
“The interesting thing about this is that there is a loop back to machine learning model for the robot to be retrained,” Boyle said. “The new labelling of the document is uploaded into the knowledge base and the model will take the changes into account.”
Overall, since it began to look at automation in 2019, NECS has built 30 robots, applied them to 49 live scalable processes, used them for over one million transactions and is now saving around 6,000 hours per year. Boyle said it had begun to look at how large language models could be used to help manage clinical documents.
These examples demonstrate the way that machine learning and AI are extending the capabilities of automation, but it is crucial to reiterate that their deployment is not all about technology. It is about the execution of work and delivery of outcomes, and freeing up people to do great things for the organisation and the citizens it serves.
Visit www.uipath.com or request a call from one of our automation experts to learn more
