Content, Featured, News, Secondary Care

Feature: Artificial Intelligence, Data & Analytics, Orion Health and WellSky


In our Artificial Intelligence, Data & Analytics feature, we speak with Orion Health and WellSky to find out more how the technologies are being adopted to support healthcare teams.

We explore how WellSky has supported the Antibiotic Review Kit (ARK) project, funded by the National Institute for Healthcare Research. We hear from Steve Reggione, Head of Operations at WellSky, who takes us through Antibiotic Usage – Data and ARK Decision Support with WellSky.

HTN also speak with Bruce Horne, Product Specialist Lead at Orion Health, and hear how the company is using machine learning, natural language processing technologies and analytics to support its customer base. We first asked Bruce a few questions:

Could you tell me how your company is embracing AI, data and analytics?

One of the latest offerings we’ve brought to the market in this area is our Machine Learning Manager. This is designed to help customers leverage the data available to them within their shared care records to gain valuable insights that can be actioned through clinical workflow to deliver enhanced patient care.

Machine Learning Manager is the product through which algorithms and calculators can be administered, deployed and monitored at scale. Our algorithm library delivers a range of locally configurable algorithms and calculators out-of-the-box to provide predictive insights in areas such as readmission risk and surgical outcomes. We can support customers to create algorithms or use the technology to administer those developed by a third party.

We also offer data de-identification and document tagging capability. We are finding that many customers are looking to unlock protected health information (PHI) for the purpose of research, but current practices for release of PHI can be time consuming, error prone and can place an organisation’s reputation at risk. The document tagger uses Natural Language Processing (NLP) to allow records to become searchable by users and surfaces data for use in reporting.

As AI and machine learning are relatively new areas for our customers, Orion Health also offers data science consultancy. This includes strategy development, maturity assessments and implementation support.

Aside from machine learning, we provide more traditional analytics through our Discover product. The nature of the solutions implemented varies based on the customer use case but it can be anything from providing structured data access to a fully-fledged dashboard set. Our out of the box dashboards deliver reporting on key operational metrics e.g. reasons for break seal events, clinical data for example reasons for A&E attendance and product adoption e.g. shared care record usage by site. Our team of deployment experts work with customers to develop bespoke dashboards, or if they have analytics capability in-house, we can provide them with the tools to build their own or simply supply access to structured data.

Can you tell me about any customer projects in this space that you’ve been involved in over the past 6 months?

The most exciting project that comes to mind is currently taking place with the New Zealand Government, where we are developing a national algorithm management solution to support scenario modelling, risk prediction, forecasting and planning in relation to the country’s pandemic response.

Development of new machine learning algorithms and models is occurring at pace, and there is a need to be able to provide a stable, automated and scalable process for operational use of approved models. Orion Health has been awarded funding via the Ministry of Business, Innovation and Employment Covid-19 Acceleration Fund to deliver this Algorithm Hub, which will provide the infrastructure, tooling and resources necessary to support operational modelling and timely information dissemination to the government, healthcare organisations and professionals.

Our project goals are to rapidly deliver a series of pandemic related models that provide immediate value, achieve broad engagement and usage across health boards, government organisations, data scientists and researchers and to lay the foundation for a national algorithm management solution that will provide value beyond the current pandemic response.

From the work that you’ve mentioned so far, are there any learnings or advice that you’d like to share?

One of the key reasons for offering data science consultancy is that we find that it can be a challenging area for healthcare organisations to break in to. There is a big learning curve and it can be overwhelming to know where to start. To use a cliché, data science is a journey, and you can only do so much of it at a time. We can support customers to develop robust use cases and from there identify the right technology and approach to deliver those requirements.

What are the two biggest challenges you are facing around AI, data and analytics at the moment?

From an analytics perspective, it can be a challenge to understand the priority reports and analysis to get in to, given that customers can feasibly report on anything once they have the data available to them. We take the time to work with our customers to establish their requirements and to help them prioritise accordingly.

Access to data can be another challenge; if you’re trying to build an analytics dashboard or machine learning algorithm that relies on certain data points, that data needs to be rich and of a good quality. Good data quality is still, as it always has been, an essential factor in being successful.

Do you think there are any downsides to the application of AI, data and analytics in healthcare at the moment?

The area in general will provide a lot of benefits to healthcare organisations. However, the adoption of technology in this area should be use case led, strategically driven and planned.  Without this approach the outputs diminish in value and result in lost resource time and end user confidence. It is essential to have the building blocks that I have mentioned, strategy, organisational readiness and quality data in place.

What’s coming up next in the realm of AI, data and analytics for Orion?

If I speak from an UK and Ireland perspective, the machine learning and data science offerings are relatively new for us, so we are actively exploring opportunities to work with early adopters and on proof of concepts.

From an analytics perspective, Discover is gaining traction with customers looking to leverage the data within their shared care records.

AI, data and analytics is an incredibly exciting area for healthcare. The team at Orion Health and I look forward to supporting customer adoption of these new technologies in the coming months and years and being a part of the shift from integrated to personalised care.

We hear from Steve Reggione, Head of Operations at WellSky, who takes us through Antibiotic Usage – Data and ARK Decision Support with WellSky:

We were made aware of the Antibiotic Review Kit (ARK) project through a variety of sources – existing customers (NHS Trusts in England, Health Boards in Wales and Scotland), NHS Digital, NHS-X, and NHS Improvement. The ARK project was funded by the National Institute for Healthcare Research to improve hospital antibiotic prescribing. Within the project a network of NHS clinicians led by Prof Martin Llewelyn of Brighton and Sussex Medical School developed and evaluated a new decision aid for prescribers. Ultimately, what brought us to Martin and the ARK team was Ann Slee, Associate CCIO (medicines) at NHS-X as their research was aligned to some of the work that WellSky was doing with NHS-X on interoperability and generally improving electronic prescribing solutions.

Over the past 40 years, WellSky has acquired about 60% of the pharmacy secondary care marketplace for managing pharmacy medicines on the WellSky medicines management platform. The majority of our customers have expressed interest in indication-based prescribing and our strategic roadmap included adding in an indication within our e-prescribing system workflow.

The research and the resulting Decision Aid tool advocating for the addition of an indication at the initial antibiotic prescription was very welcome. The research affirmed our view which sees indication as the basis for more advanced decision support. It is a key parameter of the prescribing process because the indication will influence the treatment choices the clinician will make for the diagnosis. This decision to bring the development of WellSky’s advanced decision support forward was also supported by the fact that a mandated set of indications were going to be used. The research findings and the ARK Decision Aid toolkit fitted a number of strategies.

The ARK project had clearly demonstrated that conducting an indication-based prescription was beneficial, thus being able to harness the functionality to deliver this service in a way that would meet the research team’s objectives and the wider objects of the NHS as a whole. The project also provided a framework for WellSky to build more options on and demonstrated a great opportunity for collaboration.

We brought together a focus group comprised of NHS-X colleagues, ARK project and other researchers, as well as some of WellSky’s clients, to help us consult on prototype design. Martin gave a helpful background and the specific requirements for the ARK Decision Aid toolkit. WellSky produced a prototype design and supporting documents, forming the basis of our internal specification.

Further consultations with our focus group followed during which the prototype was refined to meet not only the ARK project principles but also WellSky strategic requirements. We wanted to ensure it provided a framework for any medicine applying the same principles as in the ARK: to mandate or request that the clinician puts in the information, with the possibility to audit and subsequently generate data to either support research or inform the organisation on the outcomes for the patient.

We’ve found it very useful working with the ARK team because it has allowed us to talk to a completely different group of people who have different goals and aims, as a result of electronic prescribing. Typically, we’re talking to an end user that looks at their individual needs whereas with the ARK team were really looking to improve the prescribing process for everyone overall, relevant to antibiotic usage by capturing specific information in a standardised way.

Some of the key elements of the research certainly informed the large part of the design of the new workflow and the way in which it was going to work. The timing of when users would input certain information, what would be captured, what would be audited, what would be available for reporting purposes, those key elements were absolutely informed by the aims of the ARK Hospital project. WellSky already had some features within the system which would allow us to do what we would refer as a ‘soft stop’.

With regards to antibiotic prescribing, the review of the appropriateness of the treatment is already within our application but this is now combined with the indication [as informed by the ARK toolkit] and the system can be guided to suspend the prescription if the clinician hasn’t performed the required review action. This respects one of the key principles of the ARK i.e. a behaviour shift from “continue unless there is evidence to stop” to “stop unless there is evidence to continue”.

The new upgrade system will be included in the system of our 50 customers that use electronic prescribing. We view this sort of feature as one that all clients will benefit from. It is therefore being offered integrated into the standard application and not as an additional chargeable module. We are confident that our customers will utilise it, not least of all because we anticipate that the NHS is going to mandate that they do this in some way because of the ARK research project. The toolkit effectively provides the standard. There might be some variability in the content collected but the data captured will be generated in a standard format.

As a result, we can provide a standardised data output that other systems could interpret. It would allow us to report on antibiotic usage from our 50 customers [if permissions were there] with the related indications and all the associated audit trails. Being able to capture that information at 50 NHS organisations provides a huge additional amount of supplementary data to help support any further research that the team might want to do. It represents an important set of additional information to capture within the electronic prescribing process to help the NHS in general, improve patient outcomes, and our customers will see it as a valuable feature.

For more information on this project and Wellsky, please see: or email the team