Navigating the HTI-1 Final Rule

Updated on April 17, 2024
Medicine doctor hand working with modern computer interface as medical network concept

What EHR/EMR Vendors Need to Know About Algorithmic Transparency

The Office of the National Coordinator for Health IT (ONC) recently released the Health Data, Technology, and Interoperability (HTI-1) Final Rule, which aims to improve transparency, support information sharing, and advance interoperability in healthcare. One of the key aspects of this rule is the establishment of first-of-its-kind transparency requirements for artificial intelligence (AI) and other predictive algorithms used in certified health IT.

As EHR/EMR vendors explore the potential of Large Language Models (LLMs) for predictive decision support, they must carefully consider the implications of the HTI-1 Final Rule. LLMs are powerful AI tools that can analyze unstructured clinical data to provide insights and recommendations, but their complexity and opacity pose challenges for meeting pending transparency requirements. 

To comply with the HTI-1 Final Rule, vendors must invest in developing robust explanatory frameworks for their LLMs, prioritize ethical considerations such as minimizing bias, and integrate these tools into existing clinical workflows and user interfaces in an intuitive and user-friendly manner. By proactively addressing the technical, ethical, and usability challenges associated with LLMs, vendors can position themselves at the forefront of AI-powered predictive decision support systems while ensuring transparency, accountability, and trust in their use.

Algorithmic Transparency: A New Frontier 

The HTI-1 Final Rule introduces groundbreaking transparency requirements for AI and predictive algorithms used in certified health IT. With ONC-certified health IT supporting the care delivered by more than 96% of hospitals and 78% of office-based physicians, this regulatory approach will have far-reaching effects on the healthcare industry.

EHR/EMR vendors must now ensure that clinical users can access a consistent, baseline set of information about the algorithms they use to support decision-making. This information will help users assess algorithms for fairness, appropriateness, validity, effectiveness, and safety (FAVES).

To meet these requirements, vendors must:

  • Demonstrate the algorithms’ fairness and lack of bias.
  • Provide clear information on the algorithms’ intended use cases and limitations.
  • Document the data sources, models, and performance metrics used in algorithm development and validation.
  • Provide evidence of the algorithms’ real-world effectiveness in improving patient outcomes and clinical decision-making.
  • Implement safety monitoring and reporting systems, and provide guidance on the appropriate use and interpretation of algorithm outputs.

Vendors will need to invest significant resources in documenting, testing, and validating their algorithms, as well as developing user-friendly interfaces and educational materials. They must also establish ongoing processes for monitoring real-world performance, addressing issues, and updating user information.

The Importance of Evidence-Based Decision Support 

As the healthcare industry embraces LLMs and other AI technologies, it is essential to distinguish between evidence-based clinical decision support and predictive decision support. Evidence-based decision support, such as diagnostic prompts or out-of-range lab alerts, is generally well-established and widely accepted. These types of decision support are not the primary focus of the HTI-1 Final Rule.

However, predictive decision support, which often relies on LLMs and other AI algorithms, is the main target of the new transparency requirements. EHR/EMR vendors must be able to demonstrate the source of the training data, identify potential biases, and provide a clear path to understand how the algorithm arrived at its predictions.

Preparing for ONC Certification Criteria 

To maintain certification and meet the requirements of the HTI-1 Final Rule, EHR/EMR vendors must closely monitor the development of the certification criteria, which are expected to be released by the end of the year. These criteria will provide specific guidance on what vendors need to demonstrate to ensure compliance with the algorithmic transparency requirements.

EHR/EMR vendors should proactively assess their current and planned use of LLMs and other predictive algorithms. They must be prepared to provide detailed information on the training data, potential biases, and the decision-making process of these algorithms. Additionally, vendors should implement feedback mechanisms that allow healthcare providers to report inaccuracies or concerns about the predictive decision support tools.

Collaboration and Transparency 

As the healthcare industry navigates the new landscape of algorithmic transparency, collaboration between EHR/EMR vendors, healthcare providers, and regulatory bodies will be essential. By working together to establish best practices, share knowledge, and address potential challenges, the industry can ensure that the benefits of AI and LLMs in healthcare are realized while prioritizing patient safety and trust.

EHR/EMR vendors must embrace transparency and be proactive in addressing the requirements of the HTI-1 Final Rule. By investing in evidence-based decision support, carefully evaluating the use of LLMs, and maintaining open lines of communication with healthcare providers and regulatory bodies, vendors can position themselves for success in this new era of algorithmic transparency.

The HTI-1 Final Rule represents a significant step forward in ensuring the responsible and ethical use of AI and predictive algorithms in healthcare. As the industry continues to evolve, EHR/EMR vendors that prioritize transparency, collaboration, and patient-centered innovation will be well-prepared to navigate the challenges and opportunities that lie ahead.

Jay Blank Background2
Jay Anders, MD
Chief Medical Officer at Medicomp Systems

Jay Anders, MD is Chief Medical Officer of Medicomp Systems.