DeepMind announced in February that it was collaborating with the NHS to build an application called Streams to help hospital staff monitor patients with kidney disease and to detect early onset acute kidney injury (AKI).
The DeepMind and NHS partnership integrates the work of several institutions spanning several years. The partnership includes DeepMind’s recent acquisition of the Hark mobile app created by HELIX at the Royal College of Art at Imperial College London, as well as Streams, the early stage development mobile application created by DeepMind and NHS in conjunction with Utswo.
Until now, there’s been limited coverage around the specific data challenges, opportunities and concerns DeepMind and NHS face as they experiment and research in this area for AI potential.
An agreement uncovered by New Scientist reveals that Google and its DeepMind artificial intelligence wing have been granted access to current and historic patient data at three London hospitals run by the Royal Free NHS Trust, covering 1.6 million individuals. “This will include information about people who are HIV-positive, for instance, as well as details of drug overdoses and abortions. The agreement also includes access to patient data from the last five years,” the New Scientist reports.
The New Scientist report talks about Patient Rescue, implying that this is a broader initiative but Deep Mind have confirmed to InfoQ that this was just an old name for Streams. In addition they told us that
We are working with clinicians at the Royal Free to understand how technology can best help clinicians recognise patient deterioration - in this case acute kidney injury (AKI). We have, and will always, hold ourselves to the highest possible standards of patient data protection. This data will only ever be used for the purposes of improving healthcare and will never be linked with Google accounts or products.
DeepMind stated that data is is not allowed to leave the UK and is completely separate from any Google data. Unsurprisingly, however, concerns have been raised about the access DeepMind has to confidential patient records. A number of news outlets cite the New Scientist article and echo similar concerns around the initial scope of data capture and personal information made available to DeepMind as part of the partnership.
mathbabe.org blogger Cathy O’Neil, whose forthcoming book “ Weapons of Math Destruction” explores the emerging issues with big data, raised additional concerns to InfoQ with regards to how the model might be used. O’Neil notes,
“Once a predictive model has been trained on historical data, you can throw away the data without losing any predictive power. Moreover, the model can then be applied to a someone even without direct access to their medical history. For that reason, we should be more concerned about laws restricting the ways models are used - against us or for us - instead of merely worrying about who has access to patient data and for how long. It's the wrong question.”