Chai launches the registry of Health AI model cards

Diving briefs:

The private public partnership, formed to create standards for the adoption of artificial intelligence in healthcare, is launching a new registry for hospitals and developers to access and share information about AI tools. The hub, known as the Model Card Registry, was announced on Friday by Health AI (CHAI) and group health system members, including Providence, Cleveland Clinic and Kaiser Permanente. The goal of the registry is to make it easier for healthcare companies to assess and shop amongst their validated AI products, Chai said. According to a spokesman, AI companies and other developers can enter tools into the resist, regardless of whether they are members of Chai. It doesn't cost you to upload your model cards or access the registry, but it can include future paywall features, they said.

Dive Insights:

AI is slowly becoming more and more entrenched in the health sector. This is because providers are increasingly less dependent on algorithms, such as automating clinical notes, chatting with patients, and streamlining staff scheduling.

According to a HIMSS survey conducted in the fall, 86% of respondents are already using AI in medical institutions, and are primarily seeking administrative tasks. However, there is growing interest in more clinical use cases for AI, such as coordinating treatment plans and helping clinicians reach a diagnosis.

However, a study by HIMSS states that adoption is still an early adoption, with three healthcare providers using AI.

One problem is that hospitals and other companies interested in technology deployments have difficulty responding to the rapid rate of change in the industry and selecting a flock of tools they are looking for business. Procurement can be challenging given that health systems on the hunting of AI products must analyze technical materials and marketing presentations.

Also, providers want to be able to implement AI ethically and without patient harm, despite concerns that the healthcare industry is moving too quickly and quickly to adopt AI without proper internal or external surveillance.

Enter Chai, one of many groups trying to create a standard for responsible AI in healthcare.

The new registry in this group aims to standardize how information about AI tools standardizes the way model cards (documents on AI model development, functionality, and limitations) are co-located.

According to Chai, the model card is intended to serve the same purpose as a food nutrition label.

Registry – Chai developed it Avanadea joint venture between Microsoft and Accenture – hospitals can access model cards in one location, and manufacturers of AI tools can get information about the product in front of potential buyers.

The model card “plays an important role in the AI ​​governance process by integrating information in a easily digestible format, facilitating product comparisons, and providing standardized data to meet the needs of various stakeholders, from radiologists to legal teams,” said Elisa Bethgarwood, Associate Chief Medical Informatics Officer at Umas Memorial Health, Masachet.

Last year, Chai released a model card template for public comment. A Chai spokesperson said the feedback was varied as stakeholders wanted to remove or add information to the card depending on the industry.

Chai has yet to make any changes to the draft, but it's still not final. The group is working to check the model card h.As a suitable field for generation AI products that are more difficult to oversee than predictive models.

Chai I'm also working to understand A spokesman said how to convert a model registry into a tool that provides the right information at the right time.

“This requires in-depth testing with a variety of stakeholders and organizations, which means testing in real workflows,” the spokesman said.

Since its establishment in 2021, Chai has been established, consortium Tech Giants has grown to 1,300 member organizations, including Microsoft, Google and Amazon. Chai, which also includes members of the federal government during the Biden administration, says the goal is to create a network of quality assurance labs to evaluate AI models and develop best practices for deployment.

Chai spearheads a consortium of large tech companies, AI developers and hospitals trying to fill the gap left by federal government.

President Donald Trump overturned a Biden-era order in January to an institution that would create a schema to oversee AI, including healthcare. Instead, the Trump administration has moved to a deregulation framework aimed at prioritizing US innovation. Particularly following the shocking debut of Deepseek, a cheap, open-source Chinese foundation model that rivals the performance of the best models created in the US.

Other standard groups include trusted and responsible AI networks (trains) supported by Microsoft. Meanwhile, Nvidia, a high-tech juggernaut that builds hardware essential for AI applications, is working with clinical AI company AIDOC to create its own standard blueprint set to be released this year.