Stanford Tests Machine Learning To Manage COVID-19 Surge

Stanford University researchers are working with clinicians to see whether AI can help manage surges of COVID-19 patients, as well as identifying patients likely to need ICU care to keep them from deteriorating.

According to STAT News, the team there has already picked an existing AI tool and is posed to decide how to integrate it into its existing clinical operations infrastructure. The project is using machine learning technology to analyze patient data and score patients as to how sick they are, as well as whether they are likely to need a higher level of care.

Ron Li, a Stanford doctor and clinical informaticist leading the project, told STAT that that building the model isn’t the challenge. “It’s the workflow design, the change management, figuring out how do you develop the system the model enables,” said Li, who’s presenting the work at a virtual conference hosted by Stanford’s Institute for Human-Centered Artificial Intelligence.

The idea behind the project is to perform a quick test to find whether hospitals can integrate AI tools into their workflows. Given the demands of the coronavirus, hospitals are working to speed up AI development to a pace seldom seen in other circumstances.

The model used by Stanford is known as the Deterioration Index, and it was built and sold by Epic.

One limitation of the model is that Epic trained the model on hospital patients who didn’t have COVID-19, which might create problems in working with those who had been diagnosed with the virus, in particular given that it wasn’t designed to support care for patients it wasn’t intended to analyze.

However, the hospital already uses the Epic EHR and has integrated the Deterioration Index into its workflow, so it was an obvious choice for Li and the team.

This technology is already in use within almost 50 health systems and hundreds of hospitals. The hospitals are using the Epic model to locate hospitalized patients at the highest risk of deterioration, STAT reports.

If the Stanford team manages to validate the algorithm, it will begin using the machine learning system to support clinical steps such as prompting nurses to check in more frequently or order tests. Eventually, the system would help physicians decide what care COVID-19 patients should get, the site reports.

The project comes as hospitals embrace AI and machine learning research and using these tools as part of mission-critical programs rather than pilot efforts. According to a recent study, roughly 53% of respondents reported using machine learning and AI within their organizations, for purposes including data management attempts and embedding such technology in core enterprise applications.

About the author

Anne Zieger

Anne Zieger

Anne Zieger is a healthcare journalist who has written about the industry for 30 years. Her work has appeared in all of the leading healthcare industry publications, and she's served as editor in chief of several healthcare B2B sites.