A data feeding frenzy is happening in the NHS right now as Artificial intelligence (AI) technology companies scramble for access to NHS data.
Driven by the wide ranging potential for AI to improve healthcare – from checking laboratory results, to bed management – Artificial intelligence (AI) in the medical space has skyrocketed to the 3rd most active sector in the AI startup space.
But developing AI and machine learning products is not the same as creating more traditional technology products, you cannot just create one by writing some code. This is known as the “cold start” challenge because AI algorithms have to be trained up on masses of data before they can produce any useful insights, and like most things they are only as good as the quality of the data fed into them.
Therefore, if you want to create a high quality AI product for the healthcare market you need lots of high quality medical data. In this context the holy grail of data is the non-public patient data sets held in the UK’s NHS, as it is the largest single data source of its kind anywhere in the world. This is the reason for the data feeding frenzy, as technology companies desperate to get access to this valuable data repository are being very proactive in looking for ways to partner with the NHS.
On the face of it this is a good thing. Partnerships using NHS data to enable machine learning have led to fantastic clinical outcomes. A good example of this is the collaboration between Google’s Deep Mind and Moorfields Eye Hospital.
But I am worried. Once an AI product has been trained up on NHS data, the company that developed it can sell this product in the wider market and make a profit. There’s nothing wrong with that per se, but I am not aware of any instances of these longer term profits being shared with the NHS, without which the product may never have been developed.
So instead of short term partnerships focused solely on delivering clinical value, how can the NHS create longer term revenue from the intellectual property (IP) created by partnerships with technology companies?
In my opinion the best way to leverage this opportunity is to continue the partnerships, but ensure they are underpinned by a licensing model that agrees up front the percentage of the product’s lifetime revenue the NHS should be granted as a fair reflection of its data contribution.
This sort of model is achievable but it does require a shift in the way the NHS deals with these companies which needs to be supported by an enhancement of staff skills, or the risk is that down the line the NHS has to pay to use the new AI products it helped create.
Esther O’Sullivan is Head of Digital Strategy for BMJ. She is a specialist in impact of digital transformation on Healthcare and Academic publishing and an expert in understanding where opportunities from these transformations can be strategically applied.