Peer-Reviewed Study of 51,127 Hello Heart Participants with Hypertension Demonstrates Improved Risk Prediction Over 90 and 365 Days
[MENLO PARK, Calif. – November 19, 2024] – A peer-reviewed study published in Frontiers in Digital Health found that a novel machine learning (ML) model for prediction of heart attacks and strokes over 90- and 365-day periods outperformed traditional predictive models.
Heart attacks and strokes are both types of atherosclerotic cardiovascular disease (ASCVD), which occurs when plaque builds up in blood vessel walls and reduces blood flow to organs throughout the body. ASCVD remains the leading global cause of morbidity and mortality.
The study, led by Hello Heart researchers in collaboration with cardiologists and researchers at Sheba Medical Center and the Scripps Research Translational Institute, marks a major advancement in deploying artificial intelligence for personalized preventive strategies for cardiovascular health.
Traditional ASCVD risk tools, such as the Pooled Cohort Equations (PCEs) and PREVENT™ scores, focus on long-term risk predictions, even though research suggests that communicating shorter-term risk may better motivate patients to adopt healthier behaviors and adhere to treatment. To bridge this gap, researchers developed a model capable of short-term risk predictions utilizing real-time mobile health data, including blood pressure and heart rate, alongside clinical data from electronic health records (EHRs). The model incorporated 291 health-related features to predict short-term ASCVD risk.
The retrospective cohort study examined 51,127 adults with hypertension who enrolled in the Hello Heart cardiovascular risk self-management program between January 2015 and January 2024. Researchers found that the novel XgBoost model outperformed PCE and PREVENT in both 90-day and 365-day ASCVD risk predictions. The XgBoost model is also designed to provide ASCVD assessments for a broader hypertensive population, including those who may not have all the data required for traditional tools like PCE or PREVENT.
“These preliminary findings underscore the potential of integrating mHealth data into cardiovascular care,” said Eyal Zimlichman, MD, study co-author, Chief Transformation Officer, and Chief Innovation Officer at Sheba Medical Center. “Novel AI models that incorporate more comprehensive data show promise for expanding access to personalized care and enabling more inclusive preventive strategies.”
“This approach has the potential to advance cardiovascular care with more personalized approaches, empowering both patients and clinicians with a better understanding of short-term cardiovascular risk,” said Edo Paz, MD, study co-author and SVP, Medical Affairs at Hello Heart. “AI-driven tools like this can leverage vast amounts of data captured in EHRs, deepen insights into risk, guide clinical decision-making, and prompt personalized interventions. We look forward to continuing our research and building even more robust mechanisms for cardiovascular risk prediction and management.”
“We are only scratching the surface of what’s possible in the application of machine learning models to health data,” added Jay Pandit, MD, study co-author and Director of Digital Medicine at Scripps Research Translational Institute. “The goal now is to translate these high-performing models into actionable insights for patients and their clinicians.”
About Hello Heart
Hello Heart is on a mission to change the way people care for their hearts. The company provides the first app and connected heart monitor to help people track and manage their heart health and get real-time tips. With Hello Heart, people can take steps to control their risk of heart attacks and stroke – the leading cause of death in the United States. Peer-reviewed studies have shown that high-risk users of Hello Heart have seen meaningful drops in blood pressure, cholesterol and even weight.
Recognized as the digital leader in preventive heart health, Hello Heart is trusted by more than 130 leading Fortune 500 and government employers, national health plans, and labor organizations such as 3M, Lenovo, Northwestern Mutual, the City of Fort Worth, and the Cleveland Bakers and Teamsters Pension Fund. Hello Heart clients can save $1,676 per enrolled user annually according to an independent analysis by the Validation Institute.
Founded in 2013, Hello Heart has raised more than $138 million from top venture firms including Khosla Ventures, IVP, and Stripes. Hello Heart is a best-in-class solution on the American Heart Association’s Innovators’ Network, CVS Health Point Solutions Management offering, and many other leading health solution platforms. Visit www.helloheart.com for more information.
