Geisinger lab taps into imaging, EHR data to advance cardiology
Engineers, researchers and physician scientists are leveraging imaging and image processing technologies to improve outcomes.
A group of engineers, researchers and physician scientists at Geisinger health system are leveraging imaging and image processing technologies to improve outcomes for patients with heart disease.
Brandon Fornwalt, MD, co-director of Geisinger’s Cardiac Imaging and Technology Laboratory, says the goal of the lab is to “improve the lives of patients with cardiovascular disease through the development, implementation and evaluation of cutting-edge technologies,” including machine learning.
In addition to cardiac imaging, the lab is increasingly focused on utilizing large electronic health record and genomic databases to better understand and treat patients with genetic mutations linked to heart disease.
“With our group, we use any kind of clinical data that we can get our hands on that is going to improve our ability to understand the trajectory of a patient and how we can intervene in a positive way,” says Fornwalt. “Geisinger has access to a large amount of genomic data. Also, what makes Geisinger’s data unique is that we had an EHR system installed in 1996—so it’s been (in place) over two decades now. And we serve a very stable patient population with longitudinal data on a lot of patients.”
Last week, the Danville, Penn.-based health system announced a multi-year partnership with tech vendor Tempus to leverage artificial intelligence for the development of diagnostic and prognostic tools for cardiovascular disease. The two organizations are looking to use machine learning to predict life-threatening cardiac events so clinicians have the opportunity to intervene before it is too late.
Among the patient scenarios the Geisinger lab wants to be able to predict is which patients will be admitted or readmitted to the hospital over the short term (within 30 days to six months), as well as the ability to forecast a heart attack within the next five years, according to Fornwalt.
“You want to get ahead of things far enough so that you can do something about them,” he says. “As a physician, you can’t possibly look at all the clinical data. You can’t sift through it all—you don’t have enough time. You need a system that can summarize that data and provide actionable insights.”
Tempus’ proprietary machine learning platform will help analyze Geisinger’s vast clinical database to identify outcome and response data leading to more proactive cardiovascular treatments and interventions.
At the same time, the lab’s collaboration with Tempus has nothing to do with genomic data, Fornwalt notes.
“It’s all about the non-genomic data that physicians who treat patients with cardiovascular disease have access to,” he adds. “We chose to make the initial phase of this (partnership) about all the image, electrocardiogram and EHR data to build models that are going to help us manage patients on a population scale or help us improve the way we treat patients on an individual level.”
As an example on a population level, machine learning has tremendous potential for serving as an analytical tool to determine who “out of 12,000 patients with heart failure” need clinical intervention right now, Fornwalt contends.
“That’s a very challenging problem, and machine learning is really well cut out to start to tackle those kinds of problems because of its prediction capabilities,” concludes Fornwalt.
Brandon Fornwalt, MD, co-director of Geisinger’s Cardiac Imaging and Technology Laboratory, says the goal of the lab is to “improve the lives of patients with cardiovascular disease through the development, implementation and evaluation of cutting-edge technologies,” including machine learning.
In addition to cardiac imaging, the lab is increasingly focused on utilizing large electronic health record and genomic databases to better understand and treat patients with genetic mutations linked to heart disease.
“With our group, we use any kind of clinical data that we can get our hands on that is going to improve our ability to understand the trajectory of a patient and how we can intervene in a positive way,” says Fornwalt. “Geisinger has access to a large amount of genomic data. Also, what makes Geisinger’s data unique is that we had an EHR system installed in 1996—so it’s been (in place) over two decades now. And we serve a very stable patient population with longitudinal data on a lot of patients.”
Last week, the Danville, Penn.-based health system announced a multi-year partnership with tech vendor Tempus to leverage artificial intelligence for the development of diagnostic and prognostic tools for cardiovascular disease. The two organizations are looking to use machine learning to predict life-threatening cardiac events so clinicians have the opportunity to intervene before it is too late.
Among the patient scenarios the Geisinger lab wants to be able to predict is which patients will be admitted or readmitted to the hospital over the short term (within 30 days to six months), as well as the ability to forecast a heart attack within the next five years, according to Fornwalt.
“You want to get ahead of things far enough so that you can do something about them,” he says. “As a physician, you can’t possibly look at all the clinical data. You can’t sift through it all—you don’t have enough time. You need a system that can summarize that data and provide actionable insights.”
Tempus’ proprietary machine learning platform will help analyze Geisinger’s vast clinical database to identify outcome and response data leading to more proactive cardiovascular treatments and interventions.
At the same time, the lab’s collaboration with Tempus has nothing to do with genomic data, Fornwalt notes.
“It’s all about the non-genomic data that physicians who treat patients with cardiovascular disease have access to,” he adds. “We chose to make the initial phase of this (partnership) about all the image, electrocardiogram and EHR data to build models that are going to help us manage patients on a population scale or help us improve the way we treat patients on an individual level.”
As an example on a population level, machine learning has tremendous potential for serving as an analytical tool to determine who “out of 12,000 patients with heart failure” need clinical intervention right now, Fornwalt contends.
“That’s a very challenging problem, and machine learning is really well cut out to start to tackle those kinds of problems because of its prediction capabilities,” concludes Fornwalt.
More for you
Loading data for hdm_tax_topic #better-outcomes...