Artificial Intelligence May Help to Predict Kidney Failure

The algorithm can identify risk factors of the life-threatening disease.

Go to the profile of Scientific American
Aug 01, 2019
0
0

A computer algorithm may soon help scientists to predict and potentially prevent acute kidney injury, a life-threatening condition that impacts millions of hospital patients each year. Scientists at Google’s DeepMind Health set out to develop a computer algorithm that can pinpoint risk factors and calculate a patient’s risk for developing the disease, according to Scientific American. The team trained their algorithm using more than 700,000 anonymized health records from the US Department of Veterans Affairs. The algorithm identified more than 4,000 predictive risk factors pulled from health indicators including vital signs, test results and the use of certain prescriptions.

There are several challenges that researchers need to address before the technology can be implemented, says Joseph Vassalotti, a nephrologist and chief medical officer at the National Kidney Foundation. For example, most of the sample records belonged to male patients—only 6 percent were female—and the algorithm produced a high number of false positives. 

No comments yet.