The open-source version of the artificial intelligence Applied Model Card, which is available for download from Git Hub, is a trademark of the Coalition for Health AI. The alliance led by the medical industry announced on Thursday that it had created the card to allow AI engineers in the healthcare sector to provide crucial details about how their AI techniques are trained. According to CHAI CEO Brian Anderson, areas of the draft open-source design card – a medical AI ‘ nutrition brand ‘ – go above and beyond the U. S. Health and Human Service’s Health Data, Technology, and Interoperability: Certification Program Updates, Algorithm Transparency, and Information Sharing Final Rule for confirming wellness IT systems. The National Academy of Medicine’s AI code of conduct is one such voluntary standard that the nutrition label provides. Instead of merely leaving it up to a PowerPoint slide and anecdotal stories like to build trust, Anderson said on Wednesday,” It’s an important step in starting the conversation between a customer and a vendor.” Founded by consensus and driven by growing demand from both startups and health systems, CHAI has stated its goal is to make sure that anyone creating and using artificial intelligence ( AI ) in the health sector can make informed decisions and that the public’s access to its nutrition label increases their level of confidence and transparency. We need more transparency about how these models were created and how they perform, he said, if we want doctors, nurses, and patients to” trust” AI models that will be used in more and more consequential use cases in healthcare. Any HIT company or health system can use CHAI’s model card in any way they choose, which the coalition claimed could streamline the procurement processes and improve implementation at scale. We want the customer community and the vendor community to be both interested in having access to and use the model card widely,” Anderson said. A concerted multi-stakeholder effort to create” a consensus set of definitions about what responsible AI looks like” led to the creation of the CHAI nutrition label, Anderson explained. This includes the agreed-upon metrics for evaluation, how to think about performance, and assessments of fairness and bias. To promote AI standards, the coalition has aimed to bring together regulators and developers. When you combine vendors, AI model developers, and their customers, he said of the work of the past eight months as a “real challenge.”” When you try to reach consensus on what level of transparency we need from the developers, it is a real challenge. The coalition is asking for feedback from testing via GitHub on or before January 22 despite CHAI’s announcement in October that the certification rubric and model card designs can be anticipated after incorporating stakeholder feedback by the end of April 2025. Standards and alignmentDespite the organization’s nearly 3, 000 members, making the nutrition label freely available ultimately aids in the development of hundreds, if not thousands of AI tools. A standard that can be used again is a digital open-source version of the model card. According to Anderson, “you want to have scalable solutions that can face the difficulties of managing and monitoring” numerous AI systems and tools. Another area of study that Anderson devotes a lot of time to is alignment. For instance, many in the sector are concerned about the inclusion of patients in the development process. The CHAI AI model Card also includes a section centered on the National Academy of Medicine’s AI code of conduct, which is not included in the HTI one rule.” We think that’s really important because our patient community groups believe, and I believe developers should do that from the beginning.” Making sure vendors have the opportunity to express if they think their model’s development is in line with the AI code of conduct that NAM describes, and if they think their models should be developed in line with that is a priority, according to Andrea Fox, senior editor of Healthcare IT News.
Email: afox@himss.org
Healthcare IT News is a HIMSS Media publication.
Open-source medical AI protein label model card is available.
