The missing link in the NHS data strategy —Data scientists

A Data Science Strategy

The first priority of the NHSX draft strategy “Data saves lives: reshaping health and social care with data” is understanding “the potential for data-driven innovation”. The strategy provides 29 commitments outlining the role of data in transforming the NHS for patients, staff and innovators. It is a first step towards bridging the gap between the algorithmic successes of AI academia and the practical running of a health service.

Of the 29 commitments made in the draft strategy, three impact the health data science community.

• Building analytical and data science capability
• Encouraging AI innovation
• Supporting innovators to work with health and care organisations

Building analytical and data science capability

To build analytical and data science capability we need people who cover the kaleidoscope of skillsets that now sit under the data science umbrella. The national institute of health data science (HDRUK) are leading the charge to train 5,000 health data researchers this year. The NHS AnalystX community numbers over 15,000. The UK now has a series of well funded AI PhD programmes, and world leading healthcare AI labs — Building capability, check

Encouraging AI innovation

Encouraging AI innovation is being achieved through the development of the NHS AI lab which runs a £140m AI fund for companies to develop AI solutions. The target is 100 AI companies funded and they are well on their way with over 50 winners so far — Encouraging innovation, check.

Supporting innovators to work with health and care organisations

The NHS AI Skunkworks team, led by Giuseppe Sollazzo is a wonderful initiative. They are developing AI tools by pooling ideas from health and care organisations, prioritising the most promising and then kicking off short projects to solve them. These projects are then undertaken as a collaboration from the NHS AI lab and innovators from outside the NHS. Solutions range from reinforcement learning for optimal bed allocation to NLP approaches to creating universal data catalogues — Supporting innovators, check.

The Missing Piece

Money is being spent on private companies with the technical capabilities today and money is being spent on educating and training the healthcare data scientists of tomorrow. However, there is no commitment to create jobs and careers within the NHS for data scientists.

Healthcare data scientists and machine learning engineers are attracted to large integrated data sets like moths to a flame. The brightest light in this space is the NHS. However, in the entirety of the NHS, NHSE/I, NHSX and NHSD there are (I think) 50 data scientists. I asked a current PhD candidate on one of the above courses where people go for jobs after finishing, the answer — Pharma.

In an NHS that stopped buying fax machines in 2019 there is a huge opportunity in applying statistical and machine learning approaches within well understood local data environments. This means employing data scientists, data engineers and software engineers at a local level to understand the data, share their code and make good of the dramatic acceleration in innovation that can come from people with the technical skills working alongside domain experts.

The Problem

It is extremely difficult to hire a data scientist locally within the NHS. If ‘data saves lives’, then investing in a data career structure is as important as investing in a technology infrastructure.

At the time of writing there is one “Data Scientist” role out of the 30k or so vacancies advertised in the NHS (there are also two roles that include the word “data science”).

I spoke to an information lead for an ICS looking to hire data scientists. To try and get near to a competitive data science salary it was necessary to game the system; claim they would oversee budgets, lead teams and generally not do data science. These capabilities are clearly valued by the NHS and government, millions of pounds are going into funding private companies.

It’s time to set out how those 5000 health data researchers HDRUK will train this year can help improve health and social care from within the NHS but to do so it would be imperative that the NHS pay the going rate for these scarce skills and offer and attractive career trajectory.