Heart Test Laboratories Inc., doing business as Heartsciences, said an independent study shows its Myovista electrocardiogram (ECG) machine learning model could be a cost-effective way to predict and stratify cardiac risk.
Myovista is a 12-lead resting ECG device that is designed to also provide diagnostic information about cardiac dysfunction, typically only available via cardiac imaging. Unlike traditional ECG, which registers and displays the electrical signals of the beating heart, Myovista’s artificial intelligence (AI)-powered algorithms charts myocardial energy levels – much in the way weather maps display the energy in storms.
The study – a secondary analysis from data collected in a prospective, multicenter clinical trial – looked at Myovista’s AI-ECG algorithm as a potential tool for identifying patients subgroups in which a combination of systolic and diastolic dysfunction suggests a high-risk of major adverse cardiac events (MACE).
A total of 1,461 patients were included in the analysis: 727 in an internal cohort and 518 in a validation set who were followed over three years for MACE, rehospitalizations and cardiac death.
Results showed the deep neural network AI-ECG algorithms enabled rapid decisionmaking in an office-based setting and referral for further tests, such as echocardiography (echo), and potential interventions. Specifically, it showed the model increased the odds for flagging patients at high- vs. low-risk of MACE (21% vs. 3%, P<0.001).
Notably, the results were nearly identical to an echo-based model 21% vs. 5%, P<0.001).
The algorithm included signal-processed surface ECG (spECG) parameters to understand the electrical significance of the wavelet-transformed cardiac energy data captured by Myovista.
“The quantitative values of cardiac energy at various time points of the cardiac cycle, along with frequency and amplitude data for a specific wave, generated more than 500 features to feed machine learning algorithms,” the study’s authors write. “The eventual machine learning algorithm implemented was able to detect meaningful information from the ECG as well as spECG signals linked to the echo-derived TDA [topical data analysis] phenogroups.”
The study was published in Advocate Aurora Health’s peer-reviewed Journal of Patient-Centered Research and Reviews.
Point-of-care screening
Heartsciences hopes to make the Myovista a frontline screening tool, and alternative to conventional ECG, that can provide early warning of structural and coronary arterial heart disease.
“We expect the Myovista to help enable frontline physicians, such as primary care, to improve referral processes by identifying patients that should go to cardiology before they are at an acute disease stage or have a heart attack, and avoid unnecessary referrals – today most cardiology testing, which is expensive, has a negative outcome,” Andrew Simpson, Heartsciences’ chairman and CEO, told BioWorld in an email. “The Myovista would be a low-cost, easy to perform test in a front-line environment.”
Heartsciences won CE mark approval for Myovista in 2017. It submitted a de novo application with the U.S. FDA in 2019 and, based on feedback, is conducting a new 525-patient pivotal validation study to support resubmission.
“We are close to submitting for FDA clearance and will be classified as a new product type by the FDA,” Simpson said. In an August corporate update, the company said it anticipated FDA clearance during the current fiscal year, which ends March 31, 2023.
The South Lake, Texas-based company holds a number of U.S. and international patents related to its AI-driven ECG. In October, it announced the award of a patent from the European Patent Office for the design of a proprietary electrode and cable connectors for use with the device.
Also in October, the American Medical Association issued industry-first category III current procedural terminology codes for novel AI-assisted algorithmic ECG risk assessment for cardiac dysfunction, paving the way for reimbursement of Myovista.
In June, Heartsciences completed an initial public offering with roughly $6.38 million in gross proceeds, providing listing on Nasdaq Capital Markets under the ticker symbol HSCS.
Once its gains FDA clearance to market Myovista, Heartsciences sees a pipeline of additional use cases for its low-cost point-of-care device.
“There are a range of other algorithms, such as obstructive coronary artery disease, valvular disease, hypertrophy, myocardia that could also be possible in the fullness of time,” Simpson said.
Competition in the space
Other companies are using AI to improve detection and monitoring of cardiac disease.
Paris-based Cardiologs Technologies SAS uses AI and cloud technology to leverage medical expertise for analyzing and reporting ECG trace data. The company’s FDA-cleared and CE-marked arrhythmia diagnostic software identifies more than 20 cardiac anomalies ranging from atrioventricular block to atrial fibrillation. In November 2021, Amsterdam, the Netherlands-based Royal Philips NV snapped up Cardiologs for an undisclosed sum to expand its cardiac monitoring options.
Eko Devices Inc., of Oakland, Calif., offers an ECG-based algorithm that analyzes 15 seconds of ECG data collected with its EKO Duo digital stethoscope to screen for low ejection fraction and other heart conditions. And London-based Transformative AI Ltd. is developing AI-based software that predicts a patient’s risk of suffering sudden cardiac arrest based on subtle physiological changes.
Meanwhile, in April, Paris-based Implicity SAS secured $23 million in a series A round to support commercial expansion of its software-as-a-service platform to provide smart monitoring of connected electronic cardiac implants.
