The adoption of digital health tools has skyrocketed in recent years, especially during the COVID-19 pandemic, which fueled the exponential growth of telehealth and remote care. Additionally, the rapid advancements in artificial intelligence (AI) technologies have further fueled this digital health revolution. These tools hold tremendous potential to enhance healthcare accessibility, improve efficiency in healthcare systems, and empower patients to take charge of their health. Therefore, it is imperative to establish consensus and standards for the development, testing, deployment, and governance of these technologies, considering the proliferation of unregulated digital health tools and the growing concerns around AI regulation.
The Role of Industry Collaboration:
To address the challenges faced by the healthcare industry in navigating the world of digital health and AI, several groups have emerged to provide guidance and support. Notably, the Health AI Partnership (HAIP), consisting of healthcare organizations and ecosystem partners such as DLA Piper, released a series of practical guides in May 2023. These guides outline best practices for the safe and effective adoption of AI solutions in health systems. Additionally, the Digital Medicine Society (DiMe) organized a group of healthcare experts who proposed the Evidence DEFINED Framework, aimed at evaluating the quality of digital health tools based on clinical evidence. These initiatives demonstrate the industry’s commitment to fostering informed decision-making and standardization in the digital health landscape.
HAIP Best Practice Guides:
HAIP, led by a diverse team of clinicians, engineers, lawyers, and social scientists from esteemed institutions like Duke Health, Mayo Clinic, and UC Berkeley, conducted extensive research to develop their best practice guides. Over 90 professionals from various healthcare and related fields participated in in-depth interviews, bringing expertise in areas such as bias, ethics, regulation, and credentialing.
Based on these insights, HAIP formulated a comprehensive collection of best practice guides that cover the entire AI product life cycle in healthcare settings. These guides outline eight decision points that healthcare professionals and organizations should consider when implementing AI tools:
1. Procurement: Identifying and prioritizing problems that AI can address and assessing the feasibility and viability of AI products.
2. Development: Defining success measures and designing AI solution workflows to integrate with existing operational structures.
3. Generate evidence of safety, efficacy, and equity: Validating AI products prior to clinical use and identifying potential risks.
4. Integration: Executing AI solution rollout and managing workflow changes to prevent misuse.
5. Lifecycle management: Monitoring AI solutions over time, updating or decommissioning them as necessary.
The publication of these guides aims to establish minimum elements for the governance of AI systems in healthcare settings and empower health system leaders to make informed decisions regarding AI adoption. Feedback on the guides is invited from healthcare professionals and the wider community, allowing for continuous improvement and refinement of the practices outlined.
DiMe Evidence DEFINED Framework:
Addressing the need for standardized evaluation criteria for digital health tools, the Evidence DEFINED Framework, developed by healthcare experts organized by DiMe, focuses on assessing the clinical evidence of digital health interventions (DHIs). Existing assessment frameworks were found lacking in their ability to evaluate digital tools effectively. Consequently, the Framework emphasizes clinical evaluation, efficient review processes, and the facilitation of standardized, rigorous DHI evidence assessment.
The Framework encompasses four essential components:
1. Data privacy
2. Clinical assurance and safety
3. Usability and accessibility
4. Technical security and stability
To enable stakeholders to make informed DHI selection decisions, the Framework proposes a four-step process:
1. Screen for absolute requirements: Identify threshold requirements that potential DHIs must meet, such as regulatory compliance.
2. Apply an established evidence assessment framework: Utilize existing evaluation frameworks designed for non-digital interventions.
3. Apply the Evidence DEFINED supplementary checklist: Supplement existing frameworks with specific concerns critical to digital health tools.
4. Make actionable recommendations: Provide evidence-based recommendations for appropriate levels of DHI adoption.
The Framework acknowledges the dynamic nature of the digital health space and establishes a collaborative platform for industry feedback and updates every 6 to 12 months. While the Framework’s focus is on clinical evidence, it encourages consideration of other domains, including health equity, patient experience, cost-effectiveness, and product design.
Implications and Conclusion:
The release of HAIP’s best practice guides and DiMe’s Evidence DEFINED Framework demonstrates the industry’s growing demand for evidence-based guidance and standardization in digital health. These resources hold significant value for health systems, payers, pharmaceutical and device manufacturers, and patients alike. Inclusive collaboration and regulatory expertise remain vital for successful implementation of digital health solutions.
It is essential for organizations seeking to adopt, implement, or stay current in the digital health landscape to utilize these resources as a foundation for their decision-making processes. However, it is equally important to continually assess and adapt these recommendations to real-world contexts. DLA Piper’s Digital Health and AI practices are well-equipped to provide strategic advice on digital health adoption, supporting the operationalization of industry best practices.
For further information regarding the HAIP guides, the Evidence DEFINED Framework, or DLA Piper’s Digital Health and AI capabilities, readers can reach out to their DLA Piper relationship partner, the authors of this article, or any Healthcare industry group member.
