This site uses cookies. By continuing to browse the site you are agreeing to our use of cookies.

Find out more here.

Using machine learning to improve prediction of AKI & deterioration.

  • Research type

    Research Study

  • Full title

    Using machine learning to improve prediction of acute kidney injury and general patient deterioration.

  • IRAS ID

    193991

  • Contact name

    Back Trevor

  • Contact email

    deepmind-cro@google.com

  • Sponsor organisation

    Google DeepMind

  • Duration of Study in the UK

    2 years, 0 months, 1 days

  • Research summary

    Acute kidney injury (AKI) is a sudden and recent reduction in a person’s kidney function. In the UK 1 in 5 emergency admissions into hospital are associated with AKI, with up to 100,000 deaths each year in hospital associated with acute kidney injury. Up to 30% could be prevented with the right care. For this reason, the UK Dept of Health have said that an automated system ('national algorithm') must be put in place to alert doctors to cases of AKI.

    By combining real-time and historic electronic data that hospitals store about their patients (such as laboratory information), DeepMind have created a system which generates such alerts at the Royal Free London NHS Trust. However, it appears that the national algorithm can miss cases of AKI, can misclassify their severity, and can label some as having AKI when they don't. The problem is not with the tool which DeepMind have made, but with the algorithm itself.

    We think we can overcome these problems, and create a system which works better.

    By combining classical statistical methodology and cutting-edge machine learning algorithms (e.g. 'unsupervised and semi-supervised learning'), this research project will create improved techniques of data analysis and prediction of who may get AKI, more accurately identify cases when they occur, and better alert doctors to their presence.

  • REC name

    South Central - Oxford C Research Ethics Committee

  • REC reference

    15/SC/0693

  • Date of REC Opinion

    10 Nov 2015

  • REC opinion

    Favourable Opinion