The Coalition for Health AI (CHAI) has made a pioneering move towards the enhancement of trust in healthcare artificial intelligence (AI). This week, they released a comprehensive ‘Blueprint for Trustworthy AI Implementation Guidance and Assurance for Healthcare.’ The blueprint is the fruit of CHAI’s year-long endeavour to guide health systems, AI and data science experts, and other stakeholders in the healthcare sector towards the advancement of health AI. The document also takes a strong stand to address issues of health equity and bias.
Dr. Brian Anderson, co-founder of CHAI and Chief Digital Health Physician at MITRE, emphasizes the importance of this endeavour: “Transparency and trust in AI tools that will be influencing medical decisions is absolutely paramount for patients and clinicians.” The blueprint is designed to align health AI standards and reporting, facilitating patients and clinicians in evaluating the algorithms influencing their care.
Key Elements and Next Steps in Trustworthy AI
CHAI’s blueprint outlines key elements that define the trustworthy use of AI in healthcare. These elements include usefulness, safety, accountability and transparency, explainability and interpretability, fairness, security and resilience, and enhanced privacy. These key components ensure that the application of AI in healthcare meets the highest quality standards and fosters trust among patients and healthcare providers.
Moving beyond outlining key elements, the blueprint also provides a roadmap for the next steps to facilitate the development and use of AI tools. Central to this roadmap is health system preparedness and assessment, trustworthiness and transparency throughout an AI tool’s lifecycle, and an integrated data infrastructure to support discovery, evaluation, and assurance related to health AI.
Building on Existing Frameworks
The CHAI blueprint, while a significant milestone in itself, does not stand alone. It builds on the foundational work of the White House’s Blueprint for an AI Bill of Rights. The latter provides five guidelines for the design, use, and deployment of automated and AI-based tools. Its primary aim is to protect Americans from harm as such devices become increasingly prevalent across U.S industries, including healthcare.
Furthermore, the blueprint expands upon the guidelines set out by the U.S Department of Commerce’s National Institute of Standards and Technology (NIST) in their AI Risk Management Framework. The NIST framework is geared towards nurturing trust in AI while simultaneously mitigating risks.
The CHAI blueprint, therefore, serves as a significant stride forward in the journey towards more reliable, safe, and equitable AI in healthcare. By providing detailed guidance and building on existing frameworks, CHAI is paving the way for an era of improved trust in health AI.