Northwestern Medicine Builds Machine-Learning Model to Identify Patients With Heart Failure
Scientists and Physicians Create a Custom Tool in Epic to Help Predict Advanced Heart Failure
Heart failure, a persistent condition that can worsen over time, occurs when your heart becomes weak or stiff, impacting its ability to pump blood or relax normally. Symptoms can also be subtle and hard to identify — oftentimes mistaken for normal aging.
There are also different stages. Each year, approximately 4.5% of patients with chronic heart failure (referred to as stage C), progress to advanced or stage D heart failure.
“Heart failure is a common, costly condition affecting over 6 million US adults. Anywhere between 5% to 25% of patients with heart failure have advanced heart failure,” says Cardiologist Faraz S. Ahmad, MD, assistant professor of Cardiology and Epidemiology at Northwestern University Feinberg School of Medicine and associate director of the Northwestern Medicine Center for Artificial Intelligence at Bluhm Cardiovascular Institute .“Patients can progress subtly into advanced heart failure, and it often goes unrecognized until there is a multi-organ dysfunction. Our goal is to use AI to find these patients earlier.”
Using an innovative approach, Northwestern Medicine is working to better identify patients with heart failure before they are too sick to benefit from advanced therapies such as heart transplant and ventricular assist device (VAD) placement, which will allow physicians to intervene sooner.
Patients can progress subtly into advanced heart failure, and it often goes unrecognized... Our goal is to use AI to find these patients earlier.
Faraz S. Ahmad, MD, Assistant professor of Cardiology and Epidemiology at Northwestern University Feinberg School of Medicine
Health Record Data Helps Predict Advanced Heart Failure
In the fall of 2020, a team of scientists and physicians began creating a custom augmented intelligence-enabled workflow that embedded the output of a machine-learning model in Epic. The model sorts through millions of patient records to assess if a patient has heart failure and if that patient is progressing to advanced heart failure.
Prioritizing Patients With a Risk of Developing Heart Failure
The model’s output is embedded into an Epic workflow where the predictions are available for nurse coordinators to review, along with other elements from patient charts. If a nurse coordinator determines that a patient is an appropriate candidate for further evaluation and testing, care is coordinated with the patient’s primary cardiologist or primary care physician.
As a result of this model:
Augmenting Clinical Expertise
The project team hopes to enhance integration of the AI model into workflows by optimizing when a nurse coordinator is not needed and surfacing the prediction directly to the managing physician for referral or additional treatment considerations.
This technology demonstrates how AI can be integrated into workflows to facilitate earlier identification of patients at high risk for progressing to advanced heart failure.
The team has already begun applying the same framework to train additional models to predict the risk of developing other cardiovascular conditions, saving more lives and supporting better care.
Learn more about how Northwestern Medicine is creating other AI models, as well as supporting the latest startups — including an AI model that looks to identify aortic disease progression.
