Two dynamic analytics models developed at Johns Hopkins University predicted delirium-prone patients when tested on two datasets drawn from 100,000 stays at a Boston hospital's intensive care unit, according to new research.

WHY IT MATTERS

Delirium – sudden bouts of confusion, inattention, paranoia, agitation and hallucinations – can put patients at higher risk of prolonged hospitalization, future dementia and death. By forecasting delirium, alerted clinicians could apply countermeasures that can mitigate adverse outcomes, according to the premise of artificial intelligence research published in Anesthesiology. 

"For a lot of these physiological transitions, we think that there are early warning signs that may not be obvious to a clinician but can be picked up on using the kinds of artificial intelligence-supported pattern analysis that we used here,” Dr. Robert Stevens, associate professor of anesthesiology and critical care medicine at the Johns Hopkins University School of Medicine and senior author of the study, in announcing the new findings. 

According to the online abstract, the primary objective was to predict ICU delirium by applying machine learning to data routinely collected in electronic health records.

EHR data contains signatures that are associated with delirium risk, according to Kirby Gong, a master’s degree graduate from the Johns Hopkins Department of Biomedical Engineering and primary author of the study.

Using the publicly available ICU data, the researchers developed two predictive models. 

A static model takes a single snapshot of patient data – such as age, illness severity, other diagnoses, physiologic variables and current medications – shortly after admission to predict delirium risk at any point during a hospital stay. 

When tested with another set of ICU data from a Boston hospital, the single snapshot was able to predict which patients would get delirium 78.5% of the time.

The second dynamic model monitors information over hours and days, including repeat blood pressure, pulse and temperature readings, and continuously updates delirium risk over the coming 12 hours. When tested with the Boston ICU datasat, it predicted delirium-prone patients up to 90% of the time.

Stevens is now testing the models on Johns Hopkins Medicine ICU data and plans to design a clinical trial to test the use of the algorithms and how they could shape clinical care in patients newly admitted to intensive care, according to the announcement.

THE LARGER TREND

AI is used in precision medicine to accelerate translational and clinical research in order to advance the prediction and treatment of disease.

With machine learning, researchers are finding it's possible to predict all kinds of patient outcomes, from the effects of medication dosing to how changes in the skin barrier demonstrate allergies and autoimmune conditions, and more.

Organizations like JHU and UPMC Enterprises are looking at AI and precision medicine to improve healthcare results and streamline costs, according to healthcare leaders.

"The tangible benefits are streamlined clinical workflows, improved patient outcomes and the potential to optimize resource allocation and reduce the long-term cost of care," Dr. Matthias J, Kleinz, senior vice president and head of translational sciences at UPMC Enterprises, told Healthcare IT News last year.

ON THE RECORD

"Being able to differentiate between patients at low and high risk of delirium is incredibly important in the ICU because it enables us to devote more resources toward interventions in the high-risk population," said Stevens in the statement.

Andrea Fox is senior editor of Healthcare IT News.
Email: afox@himss.org
Healthcare IT News is a HIMSS publication.