Industry Watcher’s Digest

Buzzworthy developments of the past few days.

• Last year saw the publication of more than 22,500 peer-reviewed papers involving biomedical applications of AI. In a time of big numbers, that’s just one more. But looked at in context, it tells a compelling true tale. In 2021, the year before ChatGPT injected generative AI into the world’s technological vein, PubMed’s count in the category was (an already impressive) 12,621. The bounce was more than 75%. Even as far into the new AI age as 2021, fewer than 8,000 papers appeared. The observation is from San Fran-based Bessemer Venture Partners. In a new report, the firm calls healthcare AI an overnight success 80 years in the making. “While the hype might seem like a repeat of past waves of short-lived techno-optimism in healthcare,” the authors comment, “the data tells a different story.” 
   
• Dental and vision providers want ‘in’ on AI for relief from billing and administrative headaches. That comes clear in a report from Skygen, which supplies benefits management services to those segments of U.S. healthcare. Surveying 201 dentists and optometrists/ophthalmologists, the company found 56% exploring AI options and 35% already implementing the technology in their practices. What’s more, respondents “expressed little to no major concerns about AI integration,” collectively citing utilization management (90%) and claims adjudication (87%) as the ripest areas for AI-driven automation. 
   
• A ‘more nuanced’ understanding of value is soon to emerge. It’s rising from below as AI continues uncovering unseen patterns in unconnected data. That’s the view of two thought influencers whose take on the promise of AI in healthcare is posted in the American Journal of Managed Care. The unfurling interpretation of value set up by AI is “critical as healthcare systems worldwide grapple with escalating costs and the need for sustainable healthcare systems,” write Jason Spangler, MD, MPH, of the Center for Innovation & Value Research and John Nosta of the Nostalab think tank. It’s a quick and decent read.
   
• Deep fakes are more dangerous to healthcare than we may have thought. To comprehend the risk, consider the trusted method of checking with apparent senders of phishy email before clicking any link or attachment. That doesn’t work so well when AI can persuasively pretend to be the person you know. For another, well, this is healthcare—one of cybercriminals’ favorite targets. “Today criminals will go so far as to schedule Teams calls using impersonations,” warns Anahi Santiago, chief information security officer at ChristianaCare. “[T]hey’re on video, and they look exactly like the person you would normally engage with on video.” Santiago made the comments for Healthcare IT News and will speak at the HIMSS 2024 Healthcare Cybersecurity Forum, which is slated to hit Washington, D.C., on Halloween Day. 
   
• Whether or not AI can come up with truly new ways of thinking remains to be seen. It also remains unlikely, adds the human thinking researcher Keith Holyoak, PhD, of the psychology department at UCLA. “Machines are not truly autonomous,” says Holyoak, who also happens to be a poet. “Unprompted, a large language model does nothing at all.” Yes, but, when wielded by a human, AI—like many other technological advances—can “allow people to do things that would otherwise be impossible.” Q&A here. 
   
• That’s reassuring, Professor. But have you heard about causal AI? That’s the name they’re giving to AI that evidences the “actual ability to reason, problem-solve, understand cause and effect.” The words are from Scott Hebner, principal analyst of AI at theCUBE Research (formerly Wikibon) in Silicon Valley. He expects causal AI will mesh with small language models to form a “cohesive machine learning architecture.” Heady stuff. Hear him out in a 20-minute video posted by SiliconAnglehere. 
   
• Robot bosses make better taskmasters. When their fake minds are powered by real AI, at least. Then they’re more cost effective than their human counterparts. They make more rational decisions too. They accurately forecast your future performance. They never sleep. Need we go on? Tech writer Zac Amos fleshes out the downsides or AI bosses as well as these upsides in a piece published by Unite.AI. The upshot: “AI may not replace your boss, but it could help them do a better job.” 
   
• Recent research in the news: 
   
  • University of Illinois Chicago: Project will use AI to merge health data with input from nurses, therapists
     
  • Columbia University: Columbia nursing study unpacks views on using AI for mental healthcare
     
  • Stony Brook University: Predicting patient outcomes via machine learning models
     
  • Endocrine Society: Inaugural summit to explore artificial intelligence
     
• Healthcare AI mergers & acquisitions:
   
  • Commure completes merger with Augmedix
     
• Funding news of note:
   
  • Autonomous medical coding engine Nym announces $47M growth investment led by PSG
     
  • Rippl raises $23M in fresh Series A capital to fund multi-state expansion 
     
  • Surgical robot project at the University of Tennessee Knoxville receives up to $12M in ARPA-H funding
     
  • OpenAI has the biggest VC funding round ever—$6.6B—and crosses 250 million users in the same week
     
• From AIin.Healthcare’s news partners:
   
  • Health Imaging: AI accurately predicts breast cancer years before diagnosis
     
  • Cardiovascular Business: FDA clears new software for AI-powered CCTA assessments
     
  • Radiology Business: GE HealthCare completes $53M acquisition of AI ultrasound business