Teaching computers to read for the benefit of heart failure patients

OSF HealthCare clinicians now have an easier way to identify heart failure patients in need of certain medications and clinical interventions. The project is part of a larger effort by the OSF HealthCare Heart Failure Council to decrease hospital readmission and mortality rates.

Some patients suffering from heart failure have what’s called a reduced ejection fraction (EF) or a reduced pumping function, and show symptoms of fatigue, shortness of breath and swelling. These individuals require special medications to help their hearts function better.

It is important for physicians to identify these patients by reviewing EF values. However, the EF metric is typically buried in doctors’ notes and test results, making it difficult for other physicians to easily obtain the information they need to treat their patients with the correct medications. With the right treatment, a patient’s EF may improve, sometimes into the normal range.

The OSF HealthCare Heart Failure Council worked with the Advanced Analytics team, a part of OSF Innovation, to develop a Natural Language Processing (NLP) model that intuitively reads and pulls all EF numbers for heart failure patients into the electronic medical record system which clinicians use daily.

Building an NLP program

Physicians historically document the EF of heart failure patients in written notes and tests that are saved in a large database. The problem is that the information is mixed in with a lot of other internal and external data, so there was no easy way to find out whether a patient had been diagnosed with reduced EF in the past.

“Unfortunately, when a patient’s EF is found to be in the normal range, physicians don’t continue treating them with certain heart medications,” said Dr. Parker McRae, a change agent for Cardiovascular Quality Improvement at OSF HealthCare. “Importantly however, that person’s EF may be normal only because of the drug therapy they were receiving. Having an easily displayed historical view of a patient’s EF over time will help clinicians make better decisions on treatment.”

Working with Dr. McRae, the Advanced Analytics team developed a NLP model that was trained on a sample of more than 1,000 hand-annotated echocardiogram notes. It also uses a machine learning technique known as “named entity recognition” designed as a convolutional neural network to find EF scores. Those values are then extracted, according to rules defined by Dr. McRae and are stored in the clinical databases of OSF HealthCare.

In order to validate model performance, Dr. McRae dove into a sample of 500 echocardiograms and manually separated the EF scores from the text. The trained NLP model was then applied to the same 500 notes that Dr. McRae abstracted. The model exactly matched Dr. McRae about 97% of the time.

“The NLP model is essentially an automated and highly accurate replica of Dr. McRae’s decision-making capabilities for the specific task of abstracting EF scores from echocardiogram notes,” said Jason Weinberg, a data scientist with the Advanced Analytics team.

This validation process has been completed twice, ensuring model accuracy remains clinically relevant. The NLP model has now been applied to all OSF echocardiograms, where it can identify EF scores from over 230,000 historical notes, unlocking data that is critical to a heart failure patient’s health.

“A doctor can now pull up a patient’s medical chart and see that their EF was measured seven times over their life,” said Dr. McRae. “The provider then has a longitudinal view of the patient’s heart health and can make an informed decision on medication prescriptions.”

The NLP model is also helping OSF HealthCare build a registry of patients with reduced EF, so the analytics team can later develop a program to predict the likelihood of an individual developing heart failure. The NLP model identified almost 2,500 additional patients with low EF that were not included in the heart failure registry.  There are now 25,000 people in the heart failure registry.

A small part of a larger project

Clinicians across the Ministry now have the ability to see EF numbers when they pull up a patient’s medical record. However, the Advanced Analytics team isn’t finished training the NLP program to extract these numbers. The next phase will pull in physician notes that are on MUGA scans, cardiac MRIs and cardiac CTs to get an even more accurate EF.

The launch of the NLP program comes after the recent success of another innovative project between the OSF HealthCare Heart Failure Council and the Performance Improvement team that ensures heart failure patients receive the follow-up care they need following a hospital visit.

The council is also working to predict the likelihood of a heart failure patient to be admitted to the hospital. This will be based on normal office visits with a physician. If this information can be detected ahead of time, OSF HealthCare can provide the resources necessary to help stop the first hospital admission from happening.

Last Updated: March 21, 2019

Follow Us on Social Media

About Author: Denise Molina-Weiger

Denise Molina-Weiger is a Writing Coordinator for OSF HealthCare, where she has worked since March 2015. She initially came to OSF to write about the work taking place at the Jump Trading Simulation & Education Center, one of the world’s largest simulation and innovation centers and went on to become the Media Relations Coordinator for OSF Innovation which was developed to help the hospital system lead the way in transforming care.

Before joining the OSF HealthCare team, Denise was a reporter for Peoria Public Radio for ten years, writing on everything from politics, housing and transportation issues to hospital care in the region. She earned her bachelor’s degree in radio broadcasting from Western Illinois University in 2003 and received her master’s degree in public affairs reporting from the University of Illinois at Springfield in 2004.

Denise lives in West Peoria with her husband, son and two crazy dogs. In her spare time, she likes to snuggle on the couch with her family and watch cooking shows on Netflix. She loves taking road trips with her family and then complaining about it when they are over.

View all posts by

Tags: , , ,

Categories: Innovation