The NHS data lake: a med tech puzzle

Can data discover new medicines, cure diseases and determine the best way for individuals, groups and subgroups to be treated?

AstraZeneca’s CEO thinks so. The firm is reportedly hiring an ex-Nasa AI expert to help accelerate drug discovery. Pascal Soriot says that future capabilities and advances in drug development will be built around the collection of user data, and that the NHS is well placed to lead in this.

There has been much abstract discussion about the potential benefits of AI technology in healthcare, so it's encouraging to see a business looking at concrete applications for improving drug discovery and benefits analysis. These are notoriously complex and expensive processes, and efficiencies enabled by digital technology could soon improve the quality and quantity of the drug products available, to everyone's benefit.

An important question to ask is how data in the great data lake held by the NHS and other health providers can be collected, standardised, retrieved, stored securely, valued, audited, analysed and used to fire innovation and the next generation of discovery and development.

The complexity of addressing deep data analytics in the face of new GDPR requirements is a challenge for many organisations. When you add multiple NHS challenges of overlapping agencies, AI, funding challenges, contract workers, political division, longevity gains, digitisation of records, reorganisation, advances in medical devices and software – the risk of data 'fails' is high.

Data accuracy, security and integrity are crucial for patient outcomes, and a legal obligation under the new data protection regime. The latter in particular may not always be compatible with the accelerated advancement of medicine. Or rather, the challenge is to make it so. If practitioners, patients and researchers are to benefit from the dividends of health data, then safe and stable spaces need to be developed in which this work can take place.

A dominant idea about how to achieve this is yet to emerge, but potential approaches range from smart anonymisation to giving people legal title over their data so that they can lease it for research and development purposes.