Data Science Investment Counter —Funding raised by UK data science companies in 2018.
£ 5.640 Million

New AI from DeepMind can predict acute kidney injury before doctors

Researchers at Google’s DeepMind Health have developed an artificial intelligence (AI) able to spot symptoms of acute kidney injury (AKI) 48 hours before doctors.

The study, which was published in the journal Nature last week, describes how the AI uses inputs such as blood tests, heart rates and blood pressure to perform its calculations.

DeepMind fed the AI with data from health records of 700,000 patients, as more than 100 VA hospitals participated in the study. However, the company specified that, in order to respect patients’ privacy, no personal details such as names or social security numbers were included in the study.

“The research shows that the AI could accurately predict AKI in patients up to 48 hours earlier than it is currently diagnosed,” researchers at DeepMind wrote in a blog post

“Importantly, the model correctly predicted 9 out of 10 patients whose condition deteriorated so severely that they then required dialysis. This could provide a window in the future for earlier preventative treatment and avoid the need for more invasive procedures like kidney dialysis.”

As shown by researchers at the University of Pittsburgh, roughly 2 million people die from AKI every year, and data by the International Society of Nephrology shows that there are about 13.3 million annual cases of AKI worldwide, including more than 11 million in emerging countries.

Moreover, according to DeepMind, there is currently “no effective treatment,” as doctors and researchers “only have a vague notion as to why kidney function decreases so dramatically in many patients with acute illness or injury, or why, despite renal replacement therapy, mortality is so high.”

However, AKI can be preventable in some cases, as many as 30 per cent, if treated early.

The goal, DeepMind added in the post, is to boost patient care by using digital tools in an effort to ultimately reduce health care costs and move “from reactive to preventative models of care”.

Image via Wikipedia.


Co-working space and blog dedicated to all things data science.

Subscribe to our newsletter