Healthcare AI today: Hurry up and slow down, AI playbook, docs over bots, more

News and views you ought to know about:

• Go slow to go fast. So encourages Michael Howell, MD, MPH, chief health officer at Google Research. He’s talking about thinking deeply and governing assertively when it comes to deploying AI in healthcare. The slow but sure approach, he suggests, will keep the sector’s AI champions focused on improving patient care when—and even after—the technology builds unstoppable momentum seemingly of its own accord. 
   
  • OK, true confession, I’m veering dangerously close to putting words in Howell’s mouth here. But his quote in bold above, as applied to healthcare AI, really tickles the cranial synapses, doesn’t it? The quote comes in a 3-minute video that’s part of a comprehensive primer and update on AI in healthcare posted to JAMA Network Oct. 13. The package represents a synthesis of the best thinking to come from a JAMA summit on AI convened last fall. Its lead author is Derek Angus, MD, MPH, chair of critical care at the University of Pittsburgh and a JAMA senior editor. A sampling of essential pointers and guidance reinforcements from the article:
     
  • “[T]here is considerable enthusiasm that AI, especially given recent advances, could address long-standing challenges in the access, cost and quality of healthcare delivery. Yet the optimal path for AI development and dissemination remains unclear. In contrast to drugs or more traditional medical devices, there is little consensus or structure to ensure robust, safe, transparent and standardized evaluation, regulation, implementation and monitoring of new AI tools and technologies.”
     
  • “In theory, given the importance of AI tools’ potential effects on patients’ health outcomes, methodologically rigorous evaluations should be undertaken to generate a solid evidence base to inform their dissemination. In practice, despite wide acceptance that AI tools can have large effects on health, there is considerable debate regarding which tools require evaluation, how evaluations should be conducted and who is responsible.”
     
  • “The impact of AI on the healthcare workforce will be wide-ranging. Clinicians may be excited by the potential benefits, worry about job displacement and enjoy or resist requirements to improve their AI literacy. … The workforce composition may need to change, adding more experts in the development, implementation and evaluation of AI tools.”
     
  • “AI will massively disrupt health and healthcare delivery in the coming years. … Given the many long-standing problems in healthcare, this disruption represents an incredible opportunity. However, the odds that this disruption will improve health for all will depend heavily on creation of an ecosystem capable of rapid, efficient, robust and generalizable knowledge about the consequences of these tools on health.”
     
    • If you only read one long-form article about AI in healthcare this month, this should be it.
       
• Here’s a step-by-step guide to using AI in healthcare. It’s a work of the Boston-based Digital Medicine Society, aka “DiME.” In three easy pieces—“Identify the problem,” “Choose the right tool” and “Implement AI”—the digital medical “playbook” walks AI adopters through the basics of the journey. DiME says it published the guide to help healthcare organizations “plan, source and scale” AI technologies for real-world patient care. Check it out. 
   
• Bots better not try impersonating docs over the phone in California. Or if they do, they’d better let the hearer know they’re not the real deal. That’s because this week Gov. Gavin Newsom signed into law a bill outlawing algorithms that identify themselves, whether outright or obliquely, as physicians, nurses or any other human healthcare professionals. The bill’s passage is “a critical victory for patient safety and transparency,” Shannon Udovic-Constant, MD, president of the California Medical Association, says in prepared remarks. “In an era of rapidly advancing technology, it is essential that patients know when they are interacting with an AI system or a licensed human clinician. This bill safeguards the trust between patients and their doctors that forms the cornerstone of medicine.” 
   
• AI can lend a seriously helpful hand to pediatric anesthesiologists. Researchers found the technology can assist these vital professionals monitor oxygen levels, assess post-op pain, and size and place breathing tubes. The American Society of Anesthesiologists posted an overview Oct. 11. “Think of AI as the co-pilot, while the anesthesiologist makes all the final decisions,” explains researcher Aditya Shah, a medical student at Central Michigan University. “AI can continuously analyze thousands of data points in real time and learn patterns from past cases, spotting subtle changes sooner and helping tailor decisions to each child’s unique anatomy.” Good stuff. More here. 
   
• Eleven healthcare AI startups are about to get a big boost toward success. The shoulder to stand on will come in the form of the Mayo Clinic Platform Accelerate program. The chosen 11 will enjoy “unparallelled access” to Mayo Clinic experts in various relevant fields as well as “millions” of de-identified, longitudinal clinical records. The latter will go far in helping the startups train and validate their budding models. Heralding the crop’s arrival, Mayo Clinic Platform head John Halamka, MD, says that the “only way we can transform healthcare is by bringing together clinical experts with technology innovators.” The North Star of the accelerator enterprise, Halamka emphasizes, is to “connect innovative startups with Mayo Clinic physicians and scientists to turn breakthrough ideas into real-world healthcare solutions.” More details plus the unveiling of the 11 are here. 
   
• Among those aboard the ‘slow to go fast’ train is a multispecialty enterprise in Missouri. Jefferson City Medical Group, workplace to 125 medical professionals and 600 employees, follows a structured process for rolling out clinical AI, or any other new technology, to clinicians. The CIO and the director of IT make sure their team works closely with a committee that includes clinical representation from various specialties. Even after the committee has approved a new product for implementation, any clinician can opt out of putting it into practice. “We have healthcare professionals who say we need to use AI for patient engagement—so, for example, using AI for communicating lab results back to a patient so the provider or clinician doesn’t have to,” CIO Aaron Hendrickson tellsInformation Week. The organization knows it’s ready for any given AI use case, he adds, when its physicians say they’re ready. “There’s a lot of different use cases out there.” 
   
• Also worth a look: 
   
  • How digital tools and AI are transforming prosthetics for veterans (American Legion) 
     
  • China now leads the U.S. in this key part of the AI race (Washington Post) 
     
  • Researchers give AI failing grade in use by employees  (WorkersCompensation.com)
     
• Noteworthy research news:  
   
  • An AI system with detailed diagnostic reasoning makes its case (Harvard) 
     
  • AI models predict sepsis in children, allow preemptive care (Northwestern University)
     
  • AI predicts future X-rays to help osteoarthritis patients and their doctors see what’s coming (University of Surrey) 
     
• From AIin.Healthcare’s sibling outlets:
   
  • Radiology Business: Imaging AI firm Harrison.ai scores FDA breakthrough device designations
     
  • Radiology Business: Augmented reality startup that gives interventional radiologists ‘X-ray vision’ during procedures raises $24M