Results from Cardiosense’s SEISMIC-HF I Study of a Machine Learning Model to Non-Invasively Assess Cardiac Filling Pressure Presented as Late-Breaking Science at American Heart Association’s 2024 Scientific Sessions

• SEISMIC-HF I enrolled a large, demographically diverse population across 15 sites
• Results are an important step toward enabling widespread, non-invasive hemodynamic-guided heart failure management

CHICAGO, Nov. 20, 2024 /PRNewswire/ — Cardiosense, a medical AI company transforming the diagnosis and management of cardiac disease, announced the presentation of results from its SEISMIC-HF I study, which was designed to develop and evaluate the performance of a machine learning (ML) algorithm to non-invasively estimate pulmonary capillary wedge pressure (PCWP). Results of the SEISMIC-HF I study, a prospective, multisite, observational data collection study for further development of ML algorithms for patients with heart failure, were presented as late-breaking science by Liviu Klein, MD, at the American Heart Association’s (AHA) 2024 Scientific Sessions, and an independent review was provided by Jessica Golbus, MD, MS as discussant. 

“Effectively managing heart failure remains one of our biggest unmet clinical needs,” said Liviu Klein, MD, MS, Section Chief of Advanced Heart Failure, Mechanical Circulatory Support, Pulmonary Hypertension, and Heart Transplant at the University of California San Francisco and lead clinical advisor for Cardiosense. “Though multiple studies over the past decade have demonstrated that guiding therapy using cardiac pressures is very effective in reducing heart failure hospitalizations and mortality, measuring these pressures has traditionally required invasive procedures, limiting its utility for most patients. The results of SEISMIC-HF I are an important step towards enabling widespread hemodynamic-guided management of patients with heart failure through a non-invasive approach.”

Algorithm performance in the validation holdout set of patients with heart failure with reduced ejection fraction (HFrEF) produced an error of 1.04 ± 5.57 mmHg in estimating PCWP when compared to right heart catheterization (RHC) measurements. The ML model was trained on a large, demographically diverse population of patients with heart failure, indicating generalizable performance in a broader population for future investigation in aiding physicians in managing patients with heart failure.

“The ML model developed in SEISMIC-HF I is the first technology to non-invasively estimate intracardiac filling pressure suggesting technical accuracy similar to existing FDA-approved implantable pressure sensors in previously published studies,” said Jessica Golbus, MD, MS, assistant professor at University of Michigan Medical School and a cardiologist at the U-M Health Frankel Cardiovascular Center. “I am excited to follow the technology and the results of future studies evaluating its impact on long-term clinical outcomes, as it holds potential to improve the lives of patients with heart failure.”

“We are committed to expanding access to the best cardiology care to all patients by advancing non-invasive approaches to evaluate cardiac function,” said Amit Gupta, co-founder and CEO of Cardiosense. “Data presented at AHA this year signifies the potential of this technology as a clinically viable non-invasive option for hemodynamic-guided heart failure management. We’re excited to move one step closer towards enabling a truly proactive monitoring solution for heart failure patients across the globe.” 

For the latest news and information, follow Cardiosense on X and LinkedIn, or visit: www.cardiosense.com.

About the SEISMocardiogram In Cardiovascular Monitoring for Heart Failure I (SEISMIC-HF I) Study
SEISMIC-HF I is a prospective, multisite, observational study to collect data to develop a ML algorithm that non-invasively estimates PCWP from physiologic signals collected on the CardioTag™ device, a non-invasive wearable placed on the patient’s chest. The study was designed to enroll up to 1,000 participants scheduled to undergo a right heart catheterization (RHC) from 15 US sites. Seismocardiogram (SCG), electrocardiogram (ECG), and photoplethysmogram (PPG) waveform data from the CardioTag device and core lab-adjudicated intracardiac filling pressures from the RHC were simultaneously collected to train the ML model.

About Cardiosense
Cardiosense is a leading medical AI company redefining how we detect, monitor, and manage cardiac disease. Built on over a decade of clinical and scientific research, the company is developing novel wearable sensors and machine learning algorithms that translate raw physiological signals into clinically actionable parameters to detect early signs of cardiac disease, guide personalized therapy, and improve patient outcomes.

SOURCE Cardiosense

Go to Source