AI roundup: FDA reassures innovators, Claude challenges ChatGPT, health system goes DIY with AI, more
The FDA wants designers of goods in two health-tech categories to worry more about supplying innovative products than about pleasing the FDA.
The categories are general wellness products and clinical decision support (CDS) software. The agency issued new guidance documents, one for each of those two subjects, last week. The release was timed to coincide with the Consumer Electronics Show in Las Vegas. FDA Commissioner Marty Makary, MD, MPH, attended and shot a 1-minute video from the show floor to promote the new guidelines and, with them, the FDA’s new attitude. “There’s an AI revolution,” Makary says, adding that the updated guidance is intended to “promote more innovation with AI and medical devices.” More:
- Wearables and other general wellness products have a profound ability to identify early physiologic changes, screen for diseases and help people in their quest to be healthier. That’s straight from Makary. The new guidance specifies general wellness products may include exercise equipment, audio recordings, video games, software programs and “other products that are commonly, though not exclusively, available from retail establishments.” Products are not general wellness products when they are “intended to measure, estimate, or report physiologic values for medical or clinical purposes, including screening, diagnosis, monitoring, alerting or management of a disease or condition.”
- Physicians find clinical decision support helpful enough that the FDA needs to get out of the way as a regulator. Again, Makary’s own words. “We have a clear lane for medical-grade products,” he clarifies, “but otherwise we need to adapt with the times and be proactive with guidance so that companies and developers are not left confused about what they should be doing [with regard to] what the FDA wants.”
- In general, FDA’s guidance documents do not establish legally enforceable responsibilities. Instead, they “describe the agency’s current thinking on a topic and should be viewed only as recommendations, unless specific regulatory or statutory requirements are cited,” the CDS guidance document points out. “The use of the word should in Agency guidances means that something is suggested or recommended but not required.”
- Makary and colleagues traveled to Vegas specifically to promote AI. “We’ve been doing [just that] at the agency,” the commissioner says in the CES video. “We were the first government agency to let all employees agency-wide have access to an AI tool and to agentic AI for workflows. And we’re going to keep going.”
- Access the two new guidance documents:
- Access the two new guidance documents:
ChatGPT Health didn’t have much time to bask in its first-to-market triumph of Jan. 7. Claude for Healthcare just challenged it to a de facto duel.
Announcing the latter’s entry in the healthcare-tailored LLM competition Jan. 11, Anthropic says Claude for Healthcare can supply a robust, HIPAA-compliant medical AI experience for providers, payers and patients alike.
- Providers and payers will find several “connectors” that allow Claude for Healthcare to pull information from industry-standard systems and databases, Anthropic says. “This means clinicians and administrators can save significant time finding the data and generating the reports they need,” the company states. Claude can now connect to the CMS Coverage Database, ICD-10 and the National Provider Identifier Registry, Anthropic says. “Since HIPAA-compliant organizations can now use Claude for Enterprise,” the company notes, “they can also access existing healthcare-related connectors, including PubMed, which provides access to more than 35 million pieces of biomedical literature and allows Claude to quickly surface the latest research and produce up-to-date literature reviews.”
- Patients of clinicians using the tools will benefit by having their medical histories summarized for episode-of-care context. They’ll also get test results in plain language, be informed of AI-identified patterns in fitness and health trackers, and receive suggested questions to ask at upcoming appointments. “The aim is to make patients’ conversations with doctors more productive and to help users stay informed about their health,” Anthropic says. Consumers who subscribe to certain Claude services can also grant the bot secure access to their electronic health records.
- Detailed announcement here.
A hospital system’s homegrown AI platform is about ready for action.
Peoria, Illinois-based OSF HealthCare says its new clinical intelligence toolset, dubbed “CliniPane,” eschews alerts in favor of notices placed in front of care teams when patients’ conditions deteriorate. The platform’s algorithm does so after spotting downward trends in lab values and the like, explains OSF, a Catholic system running 17 hospitals, 200 clinical offices and two colleges of nursing.
- The guiding vision was to surface information that would help clinicians improve care without adding to their workloads or complicating their workflows. “We can integrate more data that might be able to generate insights that, from a user experience standpoint, ensure valuable [assistance] for the clinician,” says Roopa Foulger, the institution’s vice president of digital innovation. “This is almost a new blueprint that we created at OSF in how you need to build these solutions that can fit with the growing needs on the AI side and how it needs to integrate with security.”
- Get the rest from OSF HealthCare’s news operation.
Also worthwhile:
- New AI model predicts disease risk while you sleep (Stanford Medicine)
- ‘Dangerous and alarming’: Google removes some of its AI summaries after users’ health put at risk (The Guardian)
- AI is speeding into healthcare. Who should regulate it? (Harvard Gazette)
From HealthExec’s sibling news outlets:
Radiology vendor Aidoc hires former AMA president as chief medical officer (Radiology Business)
