Healthcare AI newswatch: UnitedHealth lawsuit, humanized medical AI, mobile AI care clinics, more

Buzzworthy developments of the past few days.

• A federal judge has vacated five of seven complaints against UnitedHealth Group—yet is allowing the AI-implicated suit to advance. This is the class action in which a handful of Medicare Advantage enrollees accuse the insurer of using AI to reflexively deny claims. The lawsuit, filed with a federal court in Minnesota, gives several examples of the harm allegedly done by the AI-aided claims denials. One describes a series of refusals to pay for a 74-year-old’s stroke care. The patient ended up on the hook for more than $70,000 only to die in an assisted-living facility, according to representatives of the decedent’s estate. The suit charges UnitedHealth with “bank[ing] on patients’ impaired conditions, lack of knowledge and lack of resources to appeal erroneous AI-powered decisions.” In a written ruling, Judge John Tunheim gives his rationale for dismissing the five counts and hands down his call. “The only claims that survive … are the claims for breach of contract and for breach of the implied covenant of good faith and fair dealing because the Court will only need to evaluate compliance with the insurance agreements,” Tunheim writes. “Accordingly, the Court will grant in part and deny in part UHC’s motion to dismiss, allowing Plaintiffs’ breach of contract and breach of the implied covenant of good faith and fair dealing claims to proceed.” Courthouse News Service has more. 
   
• Provider orgs ought to take a human-centric approach to clinical AI adoption. That’s humans as in healthcare workers. Or at least those who’ll use the technology. Healthcare AI developers, for their part, should create products that “ultimately serve the needs of their end users, improve their daily lives and do not negatively disrupt existing systems or ways of working.” The advice is from four professional people familiar with the field. Their guidance, posted Feb. 13 in Mondaq, calls for creating a set of “user personas” to account for the job-specific ways in which healthcare workers work. “An understanding of technology usage patterns, workflow bottlenecks and real-life stories from end-users,” the authors explain, “helps craft the persona’s behaviors, skills, needs and preferences for using technology solutions.” 
   
• CVS Health is rolling out a new app to more fully digitize consumers’ interactions with the company. The tool features AI-powered search and chat functions. In a Feb. 12 earnings call, chief strategy officer Larry McGrath named digitization as one of four strategic that will guide CVS from here on out. “We’re leveraging AI to create a more intuitive workflow and faster turnaround times to reduce frustrations for our members and provider partners,” McGrath said. “The broader demographic trends of an aging population, advances in AI and technology, the need for more personalized care and the expectations of the American public to create an improved consumer experience are opportunities where we are best positioned to lead.” CVS’s announcement of the new app is here, and the Motley Fool has the full transcript of this week’s earnings call. 
   
• Picture a nurse or physician assistant getting coached in real time by an AI agent inside a mobile clinic. Coached to do what, you ask? Oh, just patient-care things normally reserved for doctors. Now picture the action taking place in a rural area where human physicians are few and far between. The scenario makes more sense now, doesn’t it? It’s the brainchild of HHS’s Advanced Research Projects Agency for Health, or ARPA-H. The agency is funding a couple of multimillion-dollar research projects to bring the idea to fruition. At one of the sites chosen to do the hands-on work, the University of Michigan, a professor of robotics and computer science explains the plan. “We want to bring the hospital to the house or to the church parking lot,” says Jason Corso, PhD, or anyplace, really, “where the nearest medical center that performs the care the patient needs might be two hours away.” ARPA-H program overview here. 
   
• The average hospital produces 50 petabytes of data each year. The typical medical record exceeds half the word count of Hamlet. Most clinical notes are bursting with redundant information. Gastroenterologist and writer Spencer Dorn, MD, MPH, MHA, reminds us of these realities to illustrate the growing utility of clinical summaries generated quickly and quite smartly by AI. Such documentations “increasingly converge with other AI tools, most obviously AI scribes, which effectively turn conversations into notes,” Dorn points out in Forbes. “However, only one-third of clinical note text derives from conversations. Summarization tools can capture the remaining two-thirds of note text deriving from other parts of the medical record.” He has quite a bit more to say about this topic. Hear him out here. 
   
• In Scotland, some healthcare thought leaders feel the need to reassure the public that healthcare AI is nothing to fear. One is Katriona Brooksbank, PhD, of the University of Glasgow. “AI cannot operate in isolation,” she writes in a short opinion piece published by The Herald.  “Clinicians must review AI-generated results before diagnosis or treatment decisions are made, ensuring patient care remains safe and effective.” AI will not replace human clinicians, she adds, and “should instead be looked upon as a powerful, well-tested tool.”
   
• Back in the States, they’re talking about the evolution of medical decision-making into ‘an iterative dialogue’ between physicians and AI. “Where doctors once relied solely on clinical expertise and static resources, they now engage in a dynamic exchange with LLMs” (large-language AI models). It’s a dynamic that “challenges assumptions and expands thinking,” innovation theorist John Nosta observes in Psychology Today. “With medical knowledge increasing exponentially, this iterative approach helps physicians process vast amounts of information while maintaining clinical judgment.” Nosta is watching for medicine to take AI well beyond its present role as a clinical efficiency booster. Human-AI combos, he suggests, may produce nothing less than elevated human reasoning.
   
• Then too, AI will never be a cure-all for what ails U.S. healthcare. “If we are unwilling to change the problems within the health system, AI is not going to solve all those problems for us,” says Adam Sharp, MD, chief executive of Galvan, a Utah-based blockchain company focused on promoting health and wellness. How about on the level of the individual patient? “If people just think it’s going to magically improve your life,” Sharp tells a TV news operation in Salt Lake City, “I’m not sure it’s coming anytime soon.” 
   
• Recent research in the news: 
   
  • University of Southern California: AI unlocks genetic clues to personalize cancer treatment 
     
  • University of Virginia: Clinical trial explores AI-powered insulin delivery for better diabetes care
     
• Notable FDA approval activity:
   
  • FDA grants breakthrough device designation to AI model for non-small cell lung cancer
     
• Funding news of note:
   
  • Keragon, an AI-powered healthcare automation platform, secures a $7.5M seed round six months post-launch
     
• From AIin.Healthcare’s news partners:
   
  • Radiology Business: Radiologists debate the merits of AI for work list triage in emergency settings
     
  • Health Imaging: AI spots missed colorectal cancers on routine CT scans