The Department of Veterans Affairs and artificial intelligence vendor DeepMind have developed an AI system that forecasts acute kidney injury in patients.
The capability helps clinicians see potential injury early enough so clinicians can intervene to prevent deterioration.
The presence of AKI, a deadly kidney disease, is difficult to detect and can have serious consequences—such as the need for dialysis—as the conditions of patients often quickly worsen.
However, using the electronic health records of more than 700,000 patients collected from VA sites, researchers developed a deep learning approach that can predict the presence of AKI in patients as much as 48 hours earlier than it can now be diagnosed.
“Although the recognition and prompt treatment of acute kidney injury is known to be challenging, our approach may offer opportunities for identifying patients at risk within a time window that enables early treatment,” according to a study published Wednesday in the journal Nature.
Specifically, the AI model accurately predicted more than 90 percent of the most severe AKI cases for patients whose conditions deteriorated so severely that they required dialysis.
“In addition to predicting future acute kidney injury, our model provides confidence assessments and a list of the clinical features that are most salient to each prediction, alongside predicted future trajectories for clinically relevant blood tests,” state the study’s authors.
According to DeepMind, this clinical information from the model eliminates the “black box” problem often associated with AI by helping clinicians understand how the technology makes the predictions and anticipates future patient deterioration.
“Moving forward, the VA Palo Alto Health Care System in California will be exploring ways to bring these advances into clinical use,” according to the VA’s announcement. “The work leading up to this clinical trial involves complex interdisciplinary coordination to build and integrate a user-friendly platform to assist clinicians with treatment decisions.”
For reprint and licensing requests for this article, click here.