identifying advanced heart failure with machine learning
Bluhm Cardiovascular Institute developed an AI-enabled workflow to identify patients with advanced heart failure and provide them with a timely, comprehensive evaluation by a heart failure specialist. As a result of the program, hundreds of patients have been screened by a heart failure nurse coordinator and been referred to a clinic for evaluation and advanced therapies. To date, two patients who would have otherwise gone untreated have received ventricular assist devices.
Associated publication:
Augmented intelligence to identify patients with advanced heart failure in an integrated health system.
JACC: Advances |October 2022
Associated publication:
Augmented intelligence to identify patients with advanced heart failure in an integrated health system.
JACC: Advances |October 2022
Eko Digital stethoscope
Eko and Bluhm Cardiovascular Institute have been collaborating since 2019 to validate algorithms that help clinicians screen for pathologic heart murmurs and valvular heart disease during routine office visits. This collaboration was key to Eko’s development of AI-enabled digital stethoscopes that can interpret heart sounds to help screen for heart murmurs and valvular damage. Since then, the Eko Murmur Analysis Software embedded in these digital stethoscopes received Food and Drug Administration (FDA) clearance for detecting and characterizing murmurs found in adult and pediatric patients.
Associated publication:
Deep learning algorithm for automated cardiac murmur detection via a digital stethoscope platform.
Journal of the American Heart Association | May 2021
Associated publication:
Deep learning algorithm for automated cardiac murmur detection via a digital stethoscope platform.
Journal of the American Heart Association | May 2021
caption health - AI in cardiac and lung ultrasounds
Caption Health collaborated with Northwestern Medicine Bluhm Cardiovascular Institute to develop and validate their first-in-class AI-guided ultrasound capture technique, created to guide novice users without ultrasound experience on how to complete a cardiac ultrasound. The technology also automatically calculates left ventricular ejection fraction. In 2022, Caption Health received landmark FDA authorization for its Caption Guidance software, in part because of work completed at Bluhm Cardiovascular Institute.
Northwestern Memorial Hospital was the first hospital to implement Caption Guidance in clinical practice, first used in the ICUs during the COVID-19 pandemic, and now available as a diagnostic tool in the hospital, emergency department and cardiovascular intensive care units.
Bluhm Cardiovascular Institute is continuing its work with Caption Health, now partnering on a new trial investigating AI-guided ultrasound technology to complete lung ultrasounds, an underutilized therapy with a multitude of clinical applications.
In 2023, Caption Health was acquired by GE Healthcare.
Associated publications:
Utility of a deep-learning algorithm to guide novices to acquire echocardiograms for limited diagnostic use.
JAMA Cardiology | June 2021
Deep learning-based automated echocardiographic quantification of left ventricular ejection fraction: a point-of-care solution.
Circulation: Cardiovascular Imaging | June 2021
Artificial intelligence-enabled POCUS in the COVID-19 ICU: a new spin on cardiac ultrasound.
JACC: Case Report | February 2021
Northwestern Memorial Hospital was the first hospital to implement Caption Guidance in clinical practice, first used in the ICUs during the COVID-19 pandemic, and now available as a diagnostic tool in the hospital, emergency department and cardiovascular intensive care units.
Bluhm Cardiovascular Institute is continuing its work with Caption Health, now partnering on a new trial investigating AI-guided ultrasound technology to complete lung ultrasounds, an underutilized therapy with a multitude of clinical applications.
In 2023, Caption Health was acquired by GE Healthcare.
Associated publications:
Utility of a deep-learning algorithm to guide novices to acquire echocardiograms for limited diagnostic use.
JAMA Cardiology | June 2021
Deep learning-based automated echocardiographic quantification of left ventricular ejection fraction: a point-of-care solution.
Circulation: Cardiovascular Imaging | June 2021
Artificial intelligence-enabled POCUS in the COVID-19 ICU: a new spin on cardiac ultrasound.
JACC: Case Report | February 2021
heartshare - pursuing deeper understanding of heart failure with preserved ejection fraction
Sanjiv Shah, MD, director of Research for Bluhm Cardiovascular Institute, is the principal investigator of the HeartShare Data Translation Center, which received a $16.7 million award from the National Institutes of Health (NIH) to coordinate data across HeartShare program sites to study and identify new therapies for heart failure with preserved ejection fraction (HFpEF). While HFpEF affects at least 2.5 million people in the U.S., there is a limited understanding of how to prevent and effectively treat it.
Dr. Shah and investigators at Bluhm Cardiovascular Institute have pioneered the use of machine learning to apply unbiased clustering analysis using dense phenotypic data (phenomapping) to result in a novel classification of HFpEF. By conducting a unique set of clinical and phenotypic testing and evaluation, we hope to arrive at the molecular basis of HFpEF subtypes and advance corresponding targeted therapeutics.
Grant Number: U54HL160273 and U01HL160279
Associated publications:
Accelerating therapeutic discoveries for heart failure: a new public–private partnership.
Nature Reviews Drug Discovery | September 2022
Advances in machine learning approaches to heart failure with preserved ejection fraction.
Heart Failure Clinics | March 2022
Tensor factorization for precision medicine in heart failure with preserved ejection fraction.
Journal of Cardiovascular Translational Research | January 2017
Dr. Shah and investigators at Bluhm Cardiovascular Institute have pioneered the use of machine learning to apply unbiased clustering analysis using dense phenotypic data (phenomapping) to result in a novel classification of HFpEF. By conducting a unique set of clinical and phenotypic testing and evaluation, we hope to arrive at the molecular basis of HFpEF subtypes and advance corresponding targeted therapeutics.
Grant Number: U54HL160273 and U01HL160279
Associated publications:
Accelerating therapeutic discoveries for heart failure: a new public–private partnership.
Nature Reviews Drug Discovery | September 2022
Advances in machine learning approaches to heart failure with preserved ejection fraction.
Heart Failure Clinics | March 2022
Tensor factorization for precision medicine in heart failure with preserved ejection fraction.
Journal of Cardiovascular Translational Research | January 2017
react-af trial
Led by Rod S. Passman, MD, MS, director of the Center for Arrhythmia Research, the Rhythm Evaluation for AntiCoagulaTion (REACT-AF) Trial is a seven-year trial funded by a $37 million grant from The National Heart, Lung, and Blood Institute to Northwestern University and Johns Hopkins University. The study will use Apple Watches and a specially developed app available on iPhones to create personalized care for each patient. Wearable devices can potentially help end the standard “one-size-fits-all” practice of prescribing lifelong anticoagulants to people with atrial fibrillation (AFib). If proven effective, this treatment paradigm will fundamentally change the standard of care for the millions of people in the U.S. living with AFib.
medical image de-identification
As part of an effort to support tools that facilitate research at Northwestern Medicine, an open-source python package for performing medical image de-identification has been created. Using machine learning and image processing techniques medical images are de-identified and optionally prepped for further subsequent machine learning work. The package is compatible with various file formats and allows the user to specify a number of filtering and output attributes. This package is currently being implemented in the NIH funded study, HeartShare, also ongoing at Northwestern Medicine Bluhm Cardiovascular Institute.
computer vision in coronary angiograms
The SYNTAX score is a score used to determine the severity of coronary artery disease. At Northwestern Medicine we are leveraging computer vision techniques to automate this otherwise lengthy process. These techniques involve vessel segmentation and identification, noting any stenosis present in the coronary vasculature. A team from Northwestern Medicine Bluhm Cardiovascular Institute. competed at the MICCAI 2023 ARCADE grand challenge on this technique.
ultromics - precision echo analysis
Bluhm Cardiovascular Institute at Northwestern Memorial Hospital is piloting Ultromics EchoGo Core technology, which provides fully automated analysis of echocardiograms using artificial intelligence. This collaboration will help simplify key measurements taken during echocardiography studies and support timely and accurate diagnosis of various cardiovascular conditions.
testing soft, flexible wearables for cardiovascular care
In partnership with the Northwestern University Rogers Research Group, Northwestern Medicine Bluhm Cardiovascular Institute. has been exploring the role of soft, flexible, skin-integrated devices in remote patient monitoring. Examples include the deployment of multi-sensor systems using intuitive AI-based approaches for monitoring cardiovascular health across outpatient, inpatient, and post-procedural environments, and the development of novel, non-invasive sensor modalities for complex hemodynamic monitoring.
Associated publication:
Synchronized wearables for the detection of haemodynamic states via electrocardiography and multispectral photoplethysmography.
Natural Biomedical Engineering | October 2023
Associated publication:
Synchronized wearables for the detection of haemodynamic states via electrocardiography and multispectral photoplethysmography.
Natural Biomedical Engineering | October 2023
improving remote patient monitoring for heart failure with cardiosense
In a prospective clinical study, Northwestern Medicine Bluhm Cardiovascular Institute is evaluating the capability of a novel, multi-sensor device developed by Cardiosense to measure clinically important signals related to heart failure.
ai for prediction of biological age
Partnering with the Northwestern University Potocsnak Longevity Institute, Northwestern Medicine Bluhm Cardiovascular Institute is applying AI to analyze multimodal data from participants enrolled in a study of the Berne Amish and other cohorts for the study of longevity. Using diverse types of data, including epigenomic, genomic, ECG and echocardiogram, the Northwestern partners are using deep learning architectures (including transformers) and other machine learning approaches, such as contrastive learning and weakly supervised learning, to better define an individual’s biological age and create an accurate marker of aging that can be used in future clinical trials.
