VA taps Google's DeepMind to predict patient deterioration

The Department of Veterans Affairs has launched a partnership with Google’s DeepMind to analyze patient records and build a model that can predict when a patient is deteriorating.

The VA said it will allow DeepMind’s algorithm to parse through 700,000 depersonalized health records, allowing the company’s algorithm to initially focus on predicting the onset of acute kidney injuries.

“Medicine is more than treating patients’ problems,” VA Secretary David Shulkin, M.D., said in an announcement. “Clinicians need to be able to identify risks to help prevent disease. This collaboration is an opportunity to advance the quality of care for our nation’s veterans by predicting deterioration and applying interventions early.” 

The partnership is one of several ongoing efforts to modernize the VA health system. Shulkin singled out modernization as one of four goals in a six-year strategic plan (PDF). Part of that effort includes eliminating legacy systems, transitioning to cloud-based analytics and implementing a “buy-first” strategy for IT solutions.

In a report (PDF) issued last week, the agency detailed efforts by the Office of Information Technology to streamline services and use health data to focus on high-priority areas like patient care and suicide prevention.

RELATED: VA's expectations for Cerner: A ‘digital veteran’s platform,’ change management and seamless interoperability

The DeepMind partnership aligns with those broader strategic objectives, allowing the agency to initially test the predictive algorithm on a subset of patients. But DeepMind has also faced scrutiny over the company’s privacy practices in the United Kingdom. Shortly after signing a five-year deal with the National Health Service in 2016, British regulators said the partnership violated the country’s Data Protection Act by sharing data in ways patients would not have reasonably expected.

A separate report issued by DeepMind’s independent panel highlighted lack of clarity in the initial agreement with NHS and outlined several recommendations for future agreements.

DeepMind Health Clinical Lead Dominic King addressed privacy concerns in a blog post announcing the VA partnership.

“As with all of our research work, we are committed to treating the data for this project with the utmost care and respect,” he wrote. “The data being used in the research are depersonalized, meaning that any information that could be used to identify individuals has been removed before DeepMind receives it.”