For localized information and support, would you like to switch to your country-specific website for {0}?
Key takeaways
- Using AI in cardiology can support care providers to improve heart health, and has been shown to improve patient care and relieve resource pressures
- An appropriate evidence strategy is essential to support successful digital health solution adoption
- Close collaboration between developers, users, customers, and regulators is critical to advance the adoption of digital health solutions
In 2021, Lark Health collaborated with Roche Diagnostics to establish an artificial intelligence (AI)-powered lifestyle change program, known as the Lark Heart Health program. This program aimed to deliver virtual personalized coaching to help patients prevent and manage key risk factors for atherosclerotic cardiovascular disease (ASCVD) and coronary artery disease (CAD).
Heart disease is the leading cause of death in the United States, accounting for approximately 1 in every 5 deaths.1 In 2016, the prevalence of adults with cardiovascular disease (CVD) was 121.5 million people.2 In 2035, this number is projected to rise to 131.2 million (45% of the US population),3 highlighting the need for a solution that actively engages patients to help mitigate and manage their risk of CVD.
Join us as we take a deep dive into this case study with Kimberly Lockwood, Clinical Researcher for Lark Health, Priya Kulkarni, Evidence Lead for Digital Health Innovations at Roche Diagnostics, and Mathieu Chaffard, Digital Health Innovation Coordinator at Roche Diagnostics, about the use of conversational AI to help prevent heart disease.
Solving unmet needs in CVD prevention
Healthcare systems worldwide are overwhelmed. For example, more than 80% of US states lack adequate healthcare infrastructure, and 100 million Americans live in what’s called a “health professional shortage area.”4Heart disease in the US costs $229 billion between 2017 and 2018 and resulted in around 697,000 deaths in 2020.5 With the rise in prevalence of CVD, healthcare systems are under even more pressure.
The AI-powered Heart Health program
The Lark Heart Health program is an artificial intelligence (AI)-driven mobile coaching solution that provides personalized cardiovascular disease (CVD) risk counseling anytime, anywhere. Designed in accordance with guidelines from the American Heart Association (AHA), American College of Cardiology (ACC), and the National Heart, Lung, and Blood Institute (NHLBI), the Lark Heart Health program aims to provide members with the tools needed to make meaningful lifestyle changes that can help them better prevent and manage key risk factors for atherosclerotic cardiovascular disease (ASCVD) and coronary artery disease (CAD). Primary approaches include heart health-specific digital nutrition coaching, medication adherence counseling, and personalized guidance on weight loss, activity, stress, and sleep.
The Lark Heart Health program is designed to be a 12-month program for primary prevention in individuals at high risk for cardiovascular disease based on health history and behaviors, or secondary prevention for those in stable condition after a cardiovascular event. The program is intended to help participants make and maintain meaningful, evidence-based lifestyle changes and learn about their cardiovascular risk factors and acquire appropriate self-management skills.
The key features of this program include:
- Personalization: Providing real-time, personalized, on-demand coaching on weight management, nutrition, exercise, medication adherence, stress, and other factors that contribute to ASCVD risk
- Active participation: Engaging users through evidence-based conversational educational content in a way that is easy to understand and relevant to each patient
- Scalability: On-demand access to Lark coaching for users at all hours
In addition, the program gathers valuable health data that can be shared with health providers, payers, and self-insured employers to help gain insights.
Preventing cardiovascular disease: Opportunities for change
The market currently lacks an efficient, scalable tool that empowers patients to manage their own heart disease prevention, creating a clear opportunity. However, a solution’s long-term effectiveness is critically dependent on its ability to initiate and support lasting behavior change over multiple years.
Establishing a holistic approach that enables long-term behavior change to help prevent CVD events
Multiple factors impact the long-term risk of developing CVD, such as weight, high blood pressure, cholesterol, smoking, diet, and physical activity. An efficient coaching program should have a holistic approach to heart disease prevention, which we sought to test through our Heart Health pilot study.
Demonstrating the benefits of our digital health solution with robust evidence
Before launching an AI in cardiology or digital health solution, a clear regulatory assessment and comprehensive strategy for assessing pre-and post-market evidence for various stakeholders (e.g. users, payers, and regulators) is essential.
The Lark Heart Health program is designed to provide educational content, coaching, and support to help patients make lifestyle changes and adopt healthy habits that can improve their heart health and prevent future cardiovascular events.
This collaboration offered a unique opportunity to jointly define the types of evidence needed to support the value claimed by this solution.
- Assessing a digital health solution compared to traditional prescription therapies: For traditional therapies available by prescription, randomized control trials (RCTs) are used to compare a prescribed treatment to an appropriate control group. While RCTs have significant control over who participates in a study as part of the randomized sample, that structured process is not how most digital health solutions are implemented. Most consumer-facing digital health solutions are available without a prescription and may be recommended by a provider, but are not generally prescribed. In our pilot study, we aimed to enroll a broad range of patients at risk for cardiovascular disease while avoiding overly specific exclusion criteria.
- Supporting evidence generation to demonstrate return on investment (ROI): For payers considering the purchase of a clinical solution, evidence demonstrating the economic benefit of the solution is very important. Other key metrics include usability and feasibility of a solution, and how these performance indicators translate into economic benefit. This pilot study supports future evidence generation for ROI.
For more insight on what types of evidence there are for digital health solutions and for whom that evidence is important, download the white paper “Generating evidence for digital health solutions.”
AI in disease prevention: An approach for CVD
The collaboration journey between Roche Diagnostics and Lark Health began through Startup Creasphere, a global healthcare innovation platform. Experts from both sides joined forces to address a common challenge: keeping patients engaged in managing their risk factors for heart disease.
Both parties bring valuable contributions to the collaboration. For instance, Lark Health has a proprietary AI-based platform with remote patient monitoring, predictive interventions and digital behavioral modification coaching, and evidence-based quality of care education, with proven success in its chronic disease management and prevention programs.6-9 Roche Diagnostics brings medical and clinical expertise in the cardiometabolic space, a history of innovative research and development work in pharmaceuticals and diagnostics, input on Heart Health build and features, input on pilot study protocol and design, and funding for the Heart Health build and pilot study.
Two work streams were initiated: one to conceptualize the Heart Health cardiac disease prevention program and develop a minimum viable product (MVP), and the other for evidence generation strategy and pilot study.
Both workstreams worked closely together: the evidence generation work informed the coaching program development, and vice versa. As soon as it was ready, the MVP was tested out as part of a 3-month pilot study with approximately 500 enrolled participants in the US. Eligibility for the pilot study was determined using a questionnaire focused on health history and typical health behaviors, but the target population for the app is not necessarily limited to those factors. It’s important to note that many people may be at risk without even knowing it yet because CVD may develop over decades.
Pilot study outcomes
The goal of this real-world, non-interventional, single-arm, observational pilot study was to generate evidence demonstrating the feasibility and acceptability of the use of AI in cardiology through a digital cardiovascular disease prevention program. The key performance indicators below reflect areas of anticipated value for potential buyers of the Heart Health program once commercially available.
Our primary objectives at the onset of the Heart Health pilot study were as follows:
- Retention: At least 60% retention by Month 3 of the study, demonstrating that Heart Health members were still engaged in app activities after 3 months.
- Improved cardiac self-efficacy: Improvement over time in cardiac self-efficacy, reflecting the value of the Heart Health educational missions.
- Engagement with screeners: Completion of ≥ 2 screeners, demonstrating that Lark can identify events or moments that could be used to improve patient care.
- High user acceptability: User acceptability was indicated by a customer satisfaction average rating of ≥ 4 out of 5 for the overall app experience and satisfaction with the in-app ASCVD survey; At least 5 coaching conversations per month indicating high engagement.
Participants were recruited from the US via Lark’s health partners, resulting in approximately 500 enrolled participants. All four primary objectives for the pilot study were achieved.
The pilot study was completed at the end of 2022. The full study results, including a complementary analysis of additional endpoints, will be published in the coming year.
Four key learnings to apply to the development of digital health solutions
Patient activation: You should always consider how to get people engaged and confident in using your digital health solution. This is necessary if you want to create personalized strategies and gain valuable insights through key metrics. This study found an improvement in cardiac self-efficacy within two months, which, as a result, highlights the critical component of patient activation. When you engage people from the start, it will increase confidence in the given solution.
Collaboration: Collaborations between startups and larger companies have the potential to help accelerate the development and successful adoption of innovative digital health solutions. It’s important to find a good partner who can provide the resources and subject matter experts to help bring these solutions to market and drive successful adoption in the long term.
Assessment and evidence: Before developing a digital health solution, it’s important to have a clear assessment of the regulatory and payer landscape to determine the product’s purpose and intended use. Once you understand this, you need to put together a comprehensive evidence strategy to generate evidence quickly for all parties. Good quality evidence is the backbone of success for any digital solution or product and separates you from the competition. As is shown by this pilot study, this can be done relatively quickly without relying on lengthy randomized clinical trials.
Digital health solutions can increase access and positively engage users with lower health literacy: Historically, digital health studies recruit people who tend to be educated and with high socioeconomic status. From the user experience interviews, we found that individuals with lower health literacy were often particularly engaged with the program. This is a positive outcome for reaching populations who live in rural areas or may not have easy access to in-person programs for behavior change.
Altering the way we approach cardiovascular disease prevention
Coaching delivered virtually in an engaging, educational, and personal way might be an efficient way to address the current rise in CVD cases. Enabling long-term behavioral changes in patients to help them control their risk factors might also offer scalability to other chronic health conditions and a cost-effective method to release the pressure on healthcare systems globally.
Generating supporting evidence is key in broadening the acceptance of such solutions within healthcare systems, and it is too early to tell whether such solutions would significantly reduce the economic burden of related conditions in the long run.
This collaboration between Lark Health and Roche Diagnostics might pave the way towards a new approach in enabling the prevention of chronic diseases at scale, beyond just the cardiovascular field.
*Lark Health successfully participated in Startup Creasphere, a leading global digital health accelerator that connects startups with industry-leading corporate partners to drive healthcare transformation.
References
- Centers for Disease Control and Prevention. (2022). Article available from https://www.cdc.gov/heartdisease/facts.htm [Accessed May 2023]
- Benjamin et al. (2019). Circulation 139, e56–e528. Article available from https://www.heart.org/en/news/2019/01/31/cardiovascular-diseases-affect-nearly-half-of-american-adults-statistics-show [Accessed May 2023]
- American Heart Association. (2017). Report available from https://www.heart.org/-/media/Files/About-Us/Policy-Research/Fact-Sheets/Public-Health-Advocacy-and-Research/CVD-A-Costly-Burden-for-America-Projections-Through-2035.pdf [Accessed May 2023]
- Health Resources & Services Administration. (2023). Article available from https://data.hrsa.gov/topics/health-workforce/shortage-areas [Accessed May 2023]
- Centers for Disease Control and Prevention. (2022). Article available from https://www.cdc.gov/heartdisease/facts.htm [Accessed May 2023]
- Graham et al. (2022). Digit Health 8, 20552076221130619. Article availbale from https://www.ncbi.nlm.nih.gov/pmc/articles/PMC9551332/ [Accessed May 2023]
- Auster-Gussman et al. (2023). JMIR Mhealth Uhealth 1, e40865. Article available from https://mhealth.jmir.org/2023/1/e40865 [Accessed May 2023]
- Branch et al. (2022). JMIR Form Res 6, e38215. Article available from https://formative.jmir.org/2022/10/e38215 [Accessed May 2023]
- Auster-Gussman et al. (2022). Population Health Management 25, 441-8. Article available from https://www.liebertpub.com/doi/pdf/10.1089/pop.2021.0283 [Accessed May 2023]