Wearable technologies for health and fitness have been on the market for many years, but their capabilities and prevalence have exploded recently .  The average person might not consider the role these devices have in emerging health tech when buckling their smartwatch every morning, but these wearable technologies enable consumers to continuously monitor vitals, physical activity, sleep patterns and even chronic conditions. With all this data readily available, payers, providers and researchers alike, now have access to more real-time health data than ever before. The question becomes, what  should we be doing with it, and how can we be using this data for innovation and progression in healthcare? 

As artificial intelligence and predictive analytics are simultaneously growing, evolving and revolutionizing at a rapid pace, there will be novel ways to process, interpret and act on this mountain of health data.

“The perfect convergence of digital health, AI, and HEOR.”

For health economics and outcomes research, this convergence presents exciting opportunities for more granular and real-world insights into patient behavior and disease progression; augmented predictive models for resource allocation and cost-effectiveness; and most impressively, the shift from reactive medicine/treatment to preventive and personalized healthcare.  And while these innovations and opportunities present some very exciting promise, new challenges arise with data integration, privacy, bias and policy adoption.

How Are Wearables & AI Transforming Evidence Generation?

Traditionally, clinical trial data and retrospective claims analyses were the basis of HEOR decisions. Now, wearables generate real-time RWD, which enables:

• Better adherence monitoring for health interventions (e.g., tracking medication or physical activity compliance via smartwatches), i.e., are patients following the treatment regimens prescribed by their HCP?
• Continuous, real-time disease monitoring (e.g., glucose sensors for diabetes, ECG monitors for cardiac conditions) and alerts when spikes or anomalies occur.
• Refined cost-effectiveness models – Providers can incorporate daily activity, heart rate, or sleep data for more accurate quality-adjusted life years (QALYs) and cost-benefit analyses.

In this new age of HEOR, a predictive AI model using wearable data could assess long-term outcomes for patients, as well as provide a case for potential savings from early intervention.  Having this data in hand, insurers and policymakers will be better able to justify a variety of policies and affect real change for patients, sooner. Specifically, AI can have a profound positive impact on cost savings and identifying patient risk by modeling both resource allocation within the spectrum of patient risk and the financial implications of early vs. delayed treatment and intervention.

Additionally, AI could be used to design better clinical trials by identifying a precise cohort based on wearables data, making them more efficient and impactful.  Imagine for a moment that we could use the data from wearable ECG devices, providers could predict the number of hospitalizations from heart attack/strokes, allowing insurers to incentivize preventive interventions over costly emergency care.

Challenges & Considerations

‍

The potential is both exciting and limitless, but before we dive in, let's first recognize some of the potential challenges and considerations with integrating AI and wearable health tech. From data bias to privacy concerns and policy uncertainties, there is potential for notable challenges in ensuring these technologies are equitable, reliable and ethical. It is critical that stakeholders explore and address these concerns before the full value of wearable data driven insights are fully integrated into HEOR.

1. Bias 

Representative datasets are the cornerstone of good HEOR data, but traditionally, wearable tech up-take biases the younger, more tech conscious population, and potentially more divisive, higher-income populations. AI models that are trained on biased data would produce economic predictions that are inaccurate.  Furthermore, the lack of standardization in wearable sensors, not to mention user error/interferance, could affect the accuracy of the data and may limit the HEOR utility.  To combat these implications, policy-makers would need to enforce data and bias mandates; that up to now, haven’t been created.

2. Privacy & Data Ownership

While this concept does appear to show some promise, it also immediately poses some very interesting legal and ethical questions:

• Who owns wearable-generated health data? Patients? Device manufacturers? Insurers? Someone else, or a combination of these individuals?
• Ethically, should payers have the ability to use wearable data to adjust premiums? Or deny coverage?
• Will Medicare/Medicaid reimburse AI-assisted wearable monitoring?

These questions become even more complex when we consider the evolving regulatory and healthcare policy landscape. While AI is not equally regulated worldwide, there is a clear trend toward establishing frameworks that address the complex challenges posed by AI technologies.  In the U.S., the FDA put out the “Artificial Intelligence/Machine Learning (AI/ML)-Based Software as a Medical Device (SaMD) Action Plan“ in January of 2021, which outlines a 5-step initiative as a starting point for their developing frameworks for the use and regulation of AI.  Additional concerns are raised when we consider HIPAA/GDPR and how AI and wearable driven health data are to be used. Compliance and explicit consent will remain as constant hurdles, challenging the use of patient data for training AI models and further complicating how this data can be leveraged - legally and ethically.

Looking forward…

The incredible promise AI-powered wearables hold for HEOR is undeniable. There is potential for patient cost savings and improved outcomes, while fine tuning economic models using insights delivered in real time. But the challenges cannot be ignored.  Bias, policy and regulation, ethics and data privacy must be considered and addressed before we can see the full impact and benefits of this new age of HEOR.  Stakeholders for value-based care will continue to teeter totter as these landscapes evolve, AI and wearable health tech become more engrained, and policymakers and payers develop strategies for equitable and ethical data use.

In any case, the future of HEOR is sure to be exciting!

‍