Industry Watcher’s Digest

Buzzworthy developments of the past few days.

• Physicians shouldn’t think of AI as a tool to wield and sharpen. They should think of it as a medical student to mold and mentor. Up to a point, of course. The holders of the opinion support their viewpoint with several points of persuasion. For example, medical AI adds a bit of workload burden, much like a human trainee. But, like a good student, it catches on so quickly that it can “meaningfully lighten” the physician’s workload. “As AI increasingly takes on tasks that medical students currently have, we envision a concurrent shift in what medical training will look like.” The authors are a third-year medical student and an attending academic physician. MedPage Today published their thought-provoking piece July 22.
   
• Steve Jobs ‘got it’ about beautiful packaging. Medical device manufacturers can get it too—and use AI to help get it right. “It’s really important to see [medical] devices and packaging as one system, one unit,” explains subject matter expert Louisa Harvey of ClariMed in a Q&A with Healthcare Packaging. (Yes, there’s a publication dedicated to that beat.) A patient “should instantly feel a continuous energy between the device in question, the packaging and any other materials,” she adds. How can AI help with this? Find out here.
   
• HHS is looking for a few good tech leaders. Namely the agency needs a chief technology officer, a chief data officer and, yes, a chief AI officer. Announcing the openings on LinkedIn, Steven Posnack of the ONC advises interested individuals to watch for details at usajobs.gov.
   
• Healthcare insurance ops leaders want to do more with AI. But their companies have been so slow to get it going that they may be holding back the industry. The finding emerged from a survey conducted by ActiveOps. More than three-quarters of respondents, 78%, believe AI could help improve operations. Yet half the field is just starting out with the technology. Meanwhile 33% say their companies rely “too much” on assumptions and human instinct for making important operational decisions, and 30% say their decisions are “not data-driven enough.” Full report here.
   
• Healthtech supplier Commure is acquiring ambient AI medical documentation outfit Augmedix for $139 million, cash. The acquisition will allow Commure to build “the health AI operating system of the future,” the company’s CEO, Tanay Tandon, says in a news release. Augmedix’s CEO, Manny Krakaris, suggests the deal will position his company to survive and thrive over the long haul.
   
• GE Healthcare is buying Intelligent Ultrasound Group’s clinical AI software business for $51 million. The takeover will give GE control of image-analysis tools for its ultrasound portfolio, MedtechDive reports. The outlet notes Intelligent Ultrasound is selling the assets because sales growth of its clinical AI offerings has been slower than projected.
   
• Nvidia hasn’t made inroads into healthcare. Healthcare has made inroads into AI—and Nvidia has been there to help things along. The distinction is important to Nvidia development lead David Ruau, who’s involved with growing the company’s business in Europe, the Middle East and Africa. In an interview with Pharmaceutical Technology, Ruau suggests healthcare has been hankering for more GPU power to support its embrace of AI. “It’s not that we pushed into the healthcare domain for AI,” he says. “It was an emerging market, and we grew with it.”
   
• European companies can forget about using Meta’s multimodal Llama model. The Mark Zuckerberg-founded company has decided not to release the soon-to-drop model on the Old Continent because the EU’s regulatory environment has become too unpredictable. The decision “leaves a difficult situation for companies outside the EU who were hoping to provide products and services that use these models,” a watcher at The Verge points out, “as they’ll be prevented from offering them in one of the world’s largest economic markets.” Meanwhile Ars Technica tells the story from a more EU-sympathetic point of view. A senior policy reporter quotes the European Commission’s vice president for values and transparency: “In the EU, consumers are able to make truly informed choices and we now take action to safeguard this right.”
   
• Beware the epistemic risks of botsh*t. That’s the word three writers at Harvard Business Review use to encapsulate “made-up, inaccurate and untruthful chatbot content that humans uncritically use for tasks.” (They don’t replace the i with an asterisk.) “Not effectively managing the risks of botsh*t could have large consequences for individual professionals and leaders, as well as their organizations,” they warn. “Some of these risks include reputational damage, incorrect decisions, legal liability, economic losses and even human safety.” Read the whole thing.
   
• Recent research in the news:
   
  • Technical University of Munich: Are AI chatbots suitable for hospitals?
     
  • British Medical Journal: Trusted TV doctors ‘deepfaked’ to promote health scams on social media
     
  • University of Illinois: Virtual reality training for physicians aims to heal disparities in Black maternal health care
     
  • American Academy of Family Physicians: Healthcare providers weigh in on their experiences developing an AI tool to understand primary care patients’ social determinants of health
     
  • IOP Publishing launches series of open access journals dedicated to machine learning and artificial intelligence for the sciences
     
• Funding news of note:
   
  • Hummingbird Healthcare raises $10M to address persistent barriers to patient access
     
  • AI startup Cohere sees valuation soar to $5.5B after new funding round
     
• From AIin.Healthcare’s news partners:
   
  • Health Imaging: Sonio rolls out AI-based fetal ultrasound training solution
     
  • Radiology Business: AI shows potential to reduce radiologists’ chest X-ray workload in an outpatient setting
     
  • Health Imaging: Multimodal AI model helps differentiate between benign and malignant pancreatic lesions