Artificial intelligence is pushing further into healthcare on two very different fronts, and the second one shows why oversight is becoming a live issue.
On the research side, a study published in npj Digital Medicine, a Nature journal, reports that large language models—the same kind of AI behind chatbots—can forecast patients' health trajectories. According to the npj Digital Medicine paper, this capability could enable so-called "digital twins," virtual models of a patient that researchers and clinicians might use to anticipate how a person's health may unfold over time.
That promise stands in contrast to how AI is already being used in day-to-day administration. According to KUOW, federal officials have reprimanded a private company that uses AI to review Medicare claims in Washington state, citing delays in processing those claims. In other words, an AI system meant to speed up or streamline claims handling drew a formal rebuke from regulators because of how it performed in practice.
Taken together, the two items capture the moment AI in medicine is in: the technology is advancing fast enough to promise predictive, personalized care, while its real-world deployment is already running into government scrutiny when it falls short.
The sources here do not detail the company named by KUOW, the scale of the delays, or the specific clinical performance of the forecasting models, so those questions remain open.
Why it matters: as AI starts touching both how patients are treated and whether their claims get paid, the gap between what the technology promises and how it actually behaves is exactly what regulators—and patients—will be judging.