 | ● |  |
|
|
|
| | | News and views you ought to know about:- Just because the FDA has approved an AI-equipped medical device doesn’t mean the agency knows how the thing will perform in the real world. For that knowledge, FDA looks to the healthcare workers using the devices in real-world patient care. And now is the time for the workers to weigh in. Which is to say the window for publicly commenting on ways to take the measure of these devices opened Sept. 30 and will close soon—at midnight of Dec. 1, to be precise.
- FDA really needs help figuring out how clinical AI conducts itself overall as the months roll along following initial approvals of individual devices. Among the kinds of insights the agency hopes to receive: recommendations on ways to “detect, assess and mitigate performance changes over time to help assure these medical devices remain safe and effective throughout their life cycle.”
- In making the request, FDA points out healthcare AI’s performance “can be influenced by changes in clinical practice, patient demographics, data inputs and healthcare infrastructure, among other factors. Such changes, commonly referred to as data drift (or concept drift, or model drift), may lead to performance degradation, bias or reduced reliability.”
- As matters stand now, many if not most AI-enabled medical devices get sized up mainly through retrospective testing or static benchmarks. “While these methods may help establish a baseline understanding of the medical device performance, they are not designed to predict behavior in dynamic, real-world environments,” FDA explains. “Ongoing, systematic performance monitoring is increasingly recognized as relevant to maintaining safe and effective AI use by observing how systems actually behave during clinical deployment.”
- FDA underscores that it’s especially keen hear from end-users who have firsthand experience with AI tools that are a.) currently deployed at scale in real-world clinical environments, b.) supported by real-world evidence and c.) applied in clinical settings.
- Announcement here, comment submission form here.
- Health insurance claim denials: AI can be a culprit—or a catcher. Examples of alleged misusers include UnitedHealth Group, Cigna and Humana, all of which have been accused of, or even sued for, using AI to wrongfully deny claims without human review. The other side of the coin, using AI to yield fewer denials, gets aired out in the American Journal of Managed Care. “[T]he most powerful use of AI in claims is really about how you prevent errors before they even reach a payer,” a revenue cycle management expert tells the journal.
- Wait. Prevent innocent errors or thwart deliberate misdeeds? Either way, “there’s a lot that providers and health systems can do to make sure that, going back to understanding the end-to-end revenue cycle process, you embed AI or transformational tech at the very beginning, at the point of entry,” continues the RCM specialist, Clarissa Riggins of Experian Health. In so doing, she continues, “you can easily flag any missing documentation, any missing or incorrect demographic data, anything that we know or have data on that shows what the payer-specific requirements are in real time.” Read the rest.
- Don’t sit back and let AI triage patients in the ER. That’s the advice of researchers in Europe who compared AI’s prowess at the task with that of physicians and nurses. They found the algorithm hit the optimal mark only half the time. By comparison, nurses were deemed spot-on in two-thirds of cases—and physicians won the match, topping 70%. The trial used hypothetical scenarios, not real-world patient cases, which the researchers acknowledge as a limitation. Still, it was telling that the technology pulled off one surprising win when it beat the nurses at triaging the most urgent cases. This finding indicates that AI “could serve as a decision-support tool in specific clinical contexts and in overwhelmed emergency departments,” one of the study’s authors tells the European Society for Emergency Medicine. “However, excessive triaging could lead to inefficiencies, so careful integration and human oversight are crucial. Hospitals should approach AI implementation with caution and focus on training staff to critically interpret AI suggestions.”
- AI-generated phake physicians—aka deepfake doctors—are popping up in videos posted to social media. There they can perpetrate all kinds of mischief. Some are dead-ringers for actual MDs, DO’s and the like. Others are original creations. Both types can appear uncannily lifelike. “What strikes me is how convincing these synthetic doctors are, with the visuals combined with compelling but potentially fake narratives,” Ash Hopkins, PhD, of Flinders University in Australia comments to Medscape. “The potential for harm is concerning.”
- The nascent partnership over AI between the Joint Commission and the Coalition for Health AI (CHAI) has an outspoken fan in its corner. “[I]t makes sense that these two organizations would want to team up to help develop practical guidance that makes sense and speaks to the real needs, on the ground, of leaders nationwide,” enthuses Healthcare Innovation editor-in-chief Mark Hagland. “One can only wish these leaders the very best as they work to chart uncharted territory in this new world that we all find ourselves in.”
- April Koh knows mental unwellness only too well. The highly successful founder of an AI startup that shot to a $3B valuation last year first felt the pangs of clinical depression when she was just 9 years old. Her empathy for people undergoing similar struggles led her to launch the company, Spring Health, which uses AI to connect employers and providers on behalf of workers who are hurting with unseen anguish. Now Time is honoring her as one of the world’s most influential rising stars. When ChatGPT burst onto the scene in 2022, kicking off the large-language AI stampede, “suddenly people wanted to talk about AI and our technology,” Koh tells the iconic magazine. “When you address [a person’s] underlying mental-health issues, you get a thriving, more productive workforce since people call in sick less.” Full writeup here.
- Only one toolkit can advance a healthcare AI product from a tentative pilot project to an essential EMR feature. That toolkit combines rigorous evaluation systems with unassailable performance measures. For hospital executives and clinical leaders, this admixture “provides a roadmap to safe, scalable, measurable and financially responsible AI adoption,” asserts healthcare technology authority Shar Hashmi, MD, PhD, in Forbes. In the coming months and years, she adds, regulatory frameworks are likely to demand standardized AI evaluation protocols. Things could come to resemble the clinical trial phases used in drug development. “At the same time, value-based care models will push hospitals to tie key performance indicators more directly to patient outcomes, equity benchmarks and cost savings,” Hashmi predicts. “The institutions that establish disciplined AI evaluation and measurement strategies today will be the ones shaping tomorrow’s healthcare standards—and leading the next wave of innovation.”
- Noteworthy research news:
- From AIin.Healthcare’s sibling news outlets:
|
| | | |
| |
| | | How many accents do you hear in a day? Healthcare is brilliantly multicultural: clinicians, patients, families; dozens of accents in a single shift. The AI behind ambient notes has to capture all voices, not just mainstream ones. In testing, Speechmatics’ medical model reached 93% overall accuracy and 17% fewer word errors vs the next best. It keeps clinical terms intact with 96% recall at a 4% keyword error rate, even with noise and overlap. Real-time speaker separation helps you follow the conversation and the medicine. Use it live or on files. Read more here.
Is Your AI Notetaker Putting You at Risk? AI meeting tools can quietly introduce security and compliance risks when they record or store conversations without the right safeguards. In healthcare, that means potential exposure of PHI and costly governance gaps. This free Security Checklist from Fellow gives IT, Ops, and compliance leaders seven simple checks to evaluate any AI notetaker against standards like HIPAA, SOC 2, and ISO 27001. Use it to spot red flags early and ensure your organization stays protected. Before Shadow AI spreads inside your org, download the checklist to reduce your risk here.
Join Us at the Global Summit Driving Systemic Solutions to Clinician Burnout Nabla is partnering with the Ending Clinician Burnout Global Summit 2025 (Nov 6–7, virtual), a global community working together to reimagine healthcare and build lasting solutions to the clinician burnout crisis. This two-day event, co-hosted by Ending Clinician Burnout Community and Sharp Index, will bring together leading institutions like Mayo Clinic, Duke Health, Yale Medicine, Johns Hopkins, and Stanford, alongside frontline clinicians, to design systemic changes that protect clinicians and strengthen patient care. 👉 Learn more and register here: endingclinicianburnout.com
|
| | | |
|
| |
|
|
