Data science: The change the NHS needs right now
[Salford, UK] Turning data into insight and action is an aspiration across many industries, and none more so than healthcare. The challenge for the NHS has been to apply the necessary resources to the data it holds and can access, and make meaningful change that improves the quality and efficiency of the delivery of care.
Now that challenge is being addressed through initiatives across the country that are applying data science techniques to transform patient care and inform more effective service planning.
One such initiative is taking place with the data science team at the Northern Care Alliance. This NHS group is formed of global digital exemplar (GDE) Salford Royal NHS Foundation Trust and its fast follower The Pennine Acute Hospitals NHS Trust, and it is engaged in work that aims to better use of patient data.
With GDE backing, Salford Royal information and business intelligence director Emma Wright set up a new data science team, having undertaken postgraduate training in health data science to further an already-distinguished career in health informatics.
“In the NHS we are very good a looking at things through the rear-view mirror,” says Wright. Our analytics cover what has already happened, and you can’t affect change in this way. We need to focus our skills and attention on what will happen, so that we can shape services that best fit the needs of our local population.”
Using data to improve healthcare
To help achieve this change, Wright asked the team how it could use data to address some of healthcare’s major issues: preventing patient deterioration, reducing readmissions, predicting unnecessary admissions, and improving patient flow.
The results are impressive. Using observations data from 6,000 renal dialysis visits, the team can now predict the risk of patient deterioration with 94% accuracy. Using a similar size observations data set, alongside patient history it can now predict with 97% accuracy who, following discharge, will be readmitted. This can help alert doctors and nurses to those patients who may need extra care.
The team is working with the integrated health and care team on a risk stratification tool that helps predict the number of unnecessary emergency admissions. Using 24 months of data from the Salford integrated care record, and socio-economic data, it has a model in place that can predict the likelihood of admission for the next 12 months with 83% accuracy – far in excess of existing models. This is informing interventions to help reduce admissions and is being received extremely favourably by local GPs.
Using tools such as R, Azure and Python, as well as the input from clinicians, is helping the two trusts better plan and deliver care. “You can train an algorithm, but there is always a factor of human intuition. You need that,” said team member Amir Munaf.
These models and approach can also be replicated at scale across the NHS; Wright and the team will be creating blueprints as part of the GDE programme.
They are also keen to explore with other partners, such as the recently-announced Local Health and Care Record (LHCRE) programme for Greater Manchester.
National data strategy
It is also a focus for NHS Digital data lead Daniel Ray. He is working at national level to take forward a national data strategy that puts high value on data science and its counterpart artificial intelligence (AI), and looks to champion the use of data to help understand, and plan the care the NHS delivers.
“Using AI and data science can help free up doctors’ time, so that they can deliver more complex areas of care, and support more affordable and sustainable care.”
Alongside working in a cross-government collaboration on data science analytics and developing a Data Science Centre of Excellence, Ray is also looking at the policy context, seeing how best to support and regulate this growing field.
Obtaining patient consent is one potential barrier. For Ray, using the right language is imperative.
“Asking if a hospital doctor wants to use analytics as part of care is different to asking people if they want to share their data with a commercial organisation,” he says. “If we ask if a patient will share their data so they can benefit from better treatment and live longer, these are clearer benefits. We need to communicate these benefits.”
Ray is hopeful that, in a few years’ time, “the NHS has a really well articulated regulations and policy framework in the direct care space, with examples of safe implementations of data science to help deliver safer and more efficient care”.
'Data science is the change the NHS needs'
In Salford and Pennine, and in areas of the country such as Birmingham and Rotherham, those examples are coming to life.
As team member Robyn Hamilton, who is using data to understand issues around patient flow, notes: “The NHS is struggling at the moment. We can now provide insights into ways in which we can provide healthcare, more efficiently and effectively.
“Data science is the change the NHS needs right now.”
Her work, and that of her peers, is turning data to insight and action, and showing that the flowers of data-driven healthcare are blooming in today’s NHS.