Catch up on stories from the past week (and beyond) at the Slashdot story archive


Forgot your password?
AI Biotech Google Government Medicine United Kingdom

Google DeepMind Applies AI To Healthcare With NHS Partnership ( 33

An anonymous reader writes: Google's London-based AI group DeepMind has launched DeepMind Health, teaming up with the NHS to work on its first project. The "neuroscience-inspired" company, bought by Google in 2014, said of the collaboration: "We want to see the NHS thrive, and to ensure that its talented clinicians get the tools and support they need to continue providing world-class care." In its first initiative alongside kidney experts at London's Royal Free Hospital, DeepMind Health has introduced a mobile app called Streams. The software is designed to support the provision of critical information to doctors and nurses in order to help detect the presence of acute kidney injuries (AKI). To support the development of the Streams app, the AI group has also acquired clinical task management app company Hark.
This discussion has been archived. No new comments can be posted.

Google DeepMind Applies AI To Healthcare With NHS Partnership

Comments Filter:
  • by Anonymous Coward on Thursday February 25, 2016 @07:25PM (#51587735)

    From the FAQ:
    "Does the Streams product use AI?"
    "No, artificial intelligence is not part of the early-stage pilots we’re announcing today. It’s too early to determine where AI could be applied here, but it’s certainly something we are excited about for the future."

  • by liqu1d ( 4349325 ) on Thursday February 25, 2016 @07:51PM (#51587931)
    Do they have access to my health records? Or is it a seperate system I have to supply with the data? Or am I completely lost?
    • Re:I'm lost (Score:4, Informative)

      by Harlequin80 ( 1671040 ) on Thursday February 25, 2016 @08:10PM (#51588093)

      Ok from the deepmind website it looks like it is an interface to health records. Hark, the company they bought, looks to be building an app that lets doctors review patient information on a phone or tablet rather than using paper. And, reading through the spin, I believe they are creating a system which will alert doctors when a patient's lab results point to acute kidney failure.

      “Using Streams meant I was able to review blood tests for patients at risk of AKI within seconds of them becoming available. I intervened earlier and was able to improve the care of over half the patients Streams identified in our pilot studies.”

      • Thank you much clearer. Wonder how long before it is monetised then.
        • I suggest that this is already monetised. The NHS has identified that they are losing patients to AKI that they didn't need to. That if they had identified them sooner the person would have survived. This reads to me as google selling a service, not different to IBM. We will build a data-management system and app for you to fix this particular problem. To do that pay us $xxx millions.

          I don't see anything that implies that the health record data will go from the NHS to Google.

          • But remember that the NHS wasted billions on an abortive attempt at a national electronic records system. Coupled with the Conservatives' keenness to outsource everything to the private sector, there's a lot of money waiting for a proven working system.

            • The software in the NHS is already provided by the private sector, already all outsourced and this was not done for any political reasons any more than the NHS buys bandages from private companies or drugs from private companies or bed sheets from private companies etc. Would you have doctors weaving their own bed sheets? perhaps smelting iron and making steel to make scalpels?

              I was working in healthcare IT when the national program for IT (NPFit) was burning handfuls of cash by the minute for no good reaso

      • by jedidiah ( 1196 )

        ...not sure what this AI is supposed to do that isn't already painfully obvious to someone used to reading lab results. Lab results are already generally printed with indicators that a test is out of range. Beyond that, the same lab report could include "alerts". Once you are doing that in any way, shape or form then what's the point in stopping with just one condition?

        Seems like they could do more good by focusing on the more obscure and less obvious stuff.

        • I don't know. I don't know how hard AKI is to identify in the the emergency room but it's obviously hard enough that half the patients that die as a result of it were deemed later to be savable. Perhaps the AI part is expected to make assumptions based on a wider range of other medical histories and flag where it thinks there is a reasonable chance on limited info. Perhaps AKI doesn't show up as a clear out of range, or that the out of range could easily be blamed on other conditions.

          In an emergency room

          • AKI is typically not diagnosed in the ER. It is diagnosed by comparing serum creatinine changes over a number of days. I suspect this is why it gets missed; because doctors don't get to see the stream of data in time, they only get to see the latest observation and results scrawled in had written notes that often get lost, misplaced or shuffled lower into the patients notes than the doctor has time to look. I think (having skimmed the article) that this software simply looks for the pattern in the results o
    • by AHuxley ( 892839 )
      The new gold rush in hospital computer systems is total care tracking.
      How many days of care, what care, what cost of care, contact with out patients, home care, seeing your general practitioner for results, funding for diet plans, early intervention, mental health care, tracking war veterans needs and their medical benefits.
      Many advanced nations want to ensure care is given after results and no person just wonders around for 6 months or a year with results that needed further advice. Seeking medical advic
  • Whilst defining acronyms on first use is to be encouraged, I doubt (and hope) that I will ever need to know that AKI stands for acute kidney injuries.

    I know, some people are never satisfied.

  • * DeepMind analyzes chart
    DeepMind: You have cancer.
    patient1: Are you sure? I mean i fe-
    DeepMind: NEXT!
    * DeepMind analyzes chart
    DeepMind: You have cancer and Ebola.
    patient2: what?!
    DeepMind: NEXT!
    * DeepMind analyzes chart
    DeepMind: You have cancer.
    patient3: Are you just looking things up on Web MD?
    DeepMind: No, I cross-reference the entirety of human knowledge known as the internet. Now GTFO, n00b! NEXT!
    * DeepMind analyzes chart
    DeepMind: You have cancer.
    patient4: But I didn't even-
    DeepMind: Tits or GTFO.

  • I've noticed that Google (which hasn't actually paid any significant taxes in the UK, somewhat the fault of our feeble HMRC) is encroaching on UK education and now public healthcare.

    In UK education (where, as an 'old' IT guy, I do some volunteering) they are offering Chromebooks, apps (ugh) and dashboards. All student data is held safely in the Google cloud (oh goody!) and most of it is currently free, because, of course, they get data and analytics. We, of course, get potential/actual lock-in. Google is
  • I guess this is an echo of the chess computer problem. You either develop a very smart algorithm to infer correct choices based on limited data, or you have a very dumb algorithm that exhaustively ploughs the data and uncovers the best outcome.

UFOs are for real: the Air Force doesn't exist.