DeepMind is the artificial intelligence company that is owned by Alphabet. It recently unveiled a new research report where it details the development of technology that can predict acute kidney injuries in patients about two days earlier than what is possible today.
DeepMind, in collaboration with the US Department of Veteran Affairs, was able to apply AI to de-identified electronic health record database that obtained from VA medical facilities. The database featured 703,782 adult patients over 172 inpatient and 1,062 outpatient sites. The research was published in Nature and showed that the system was able to accurately predict acute kidney injury 48 hours earlier as opposed to current methods. It was also able to predict nine out of ten patients that would actually need treatment such as kidney dialysis owing to extreme deterioration.
Furthermore, the model was able to predict 55.8% of all inpatient episodes of acute kidney injury and 90.2% of all acute kidney injuries that needed enhanced invasive treatment. DeepMind has also shared the results of a peer review of its mobile medical assistant called Streams. Streams has been used at the Royal Free London NHS Foundation Trust ever since the start of 2017. The app relies on an AKI algorithm for detection of patient deterioration and is capable of accessing the medical information at the bedside while communicating spontaneously with clinical terms.
The Royal Free has said that Streams allowed it to save up to two hours every day with specialists reviewing important cases in under fifteen minutes or less. The process, without Streams, generally took several hours. Very few cases of acute kidney injury were missed – the rate dropped to 3.3% instead of 12.4%. According to DeepMind, the average cost of admission per patient also dropped by 17%.
Mustafa Suleyman, co-founder & head of applied AI and Dominic King, health lead, at DeepMind said, ‘Over the last few years, our team at DeepMind has focused on finding an answer to the complex problem of avoidable patient harm, building digital tools that can spot serious conditions earlier and helping doctors and nurses deliver faster, better care to patients in need. This is our team’s biggest healthcare research breakthrough to date, demonstrating the ability to not only spot deterioration more effectively but actually predict it before it happens.’