 | ● |  |
|
|
|
| | | News and views you ought to know about:- Getting healthcare AI to fulfill its destiny—can we call it that now?—will require ‘scenario-based’ regulation of every use case. The evaluations need to look for potential lapses in ethical deployments as well as safety measures. That’s the view of attorneys with the newish global law firm A&O Shearman. What are they talking about here? Essentially, a risk-based framework for regulating AI in healthcare. “This approach involves varying the approval processes based on the risk level of each application,” write bloggers with the big firm, which formed in 2024 upon the merger of two large practices. “Essentially, the higher the risks associated with the AI tools, the more controls and safeguards should be required by authorities.”
- The authors name as examples of low-risk AI deployments those that conduct medical training, promote disease awareness and perform medical automation. Higher-risk: AI tools that perform autonomous surgery and critical monitoring. “By tailoring the regulatory requirements to the specific risks,” the bloggers contend, “we can foster innovation while ensuring that safety is adequately protected.”
- The attorneys also urge review teams to include representatives from every possible stakeholder group. “[T]his may include professionals with backgrounds in healthcare, medical technology, legal and compliance, cybersecurity, ethics and other relevant fields as well as patient interest groups,” they write. “By bringing together diverse perspectives, the complexities and ethical considerations of AI in healthcare can be better addressed, fostering trust and accountability.”
- Read the whole thing.
- Stanford AI experts agree: The tech ethos of moving fast and breaking things isn’t healthy for healthcare. In a paper published in NEJM AI, senior author Jonathan Chen, MD, PhD, and colleagues concentrate on the particular risks raised by agentic AI. “Agent-based task frameworks and benchmarks are the necessary next step to advance the potential and capabilities for effectively improving and integrating AI systems into clinical workflows,” they write. To help model developers along this next step, the Stanford AI experts have come up with a benchmarking toolkit—MedAgentBench—and made it publicly available online.
- “Our evaluation of state-of-the-art LLMs reveals that, although they demonstrate promising capabilities, they are not yet capable of reliably handling the full complexity of clinically relevant tasks,” Chen and co-authors warn. “This underscores the critical need for further optimization and iteration, positioning MedAgentBench as a pivotal benchmark to drive innovation and guide the development of agentic AI systems suitable for real-world clinical integration.”
- There’s more on the work from Stanford’s Human-Centered Artificial Intelligence center, aka “HAI,” here.
- Six states are piloting CMS’s AI-centric program for reducing wasteful federal spending on inappropriate medical services. Ohio being one of them, Cleveland.com takes a look at what’s involved. The outlet homes in on fears about CMS’s planned use of AI for making prior authorization decisions. The program “creates a barrier between what physicians and other healthcare providers order and want as medically necessary for their patients and what can be provided based on algorithms,” says the founder of a Medicare watchdog. “It will save money at the cost of the patients,” adds the executive director of a patient advocacy group. And it’s “a backdoor way of putting everybody in a Medicare Advantage plan,” says a Georgetown professor with expertise in Medicare policy. “It’s a first step to getting rid of, or downgrading, the freedom that traditional Medicare provides.”
- For its part, CMS says the AI-toting program—introduced this summer as WISeR, for Wasteful and Inappropriate Service Reduction—will help patients and providers avoid unnecessary healthcare procedures while safeguarding federal taxpayer dollars. “CMS is committed to crushing fraud, waste and abuse, and the WISeR model will help root out waste in Original Medicare,” CMS Administrator Mehmet Oz, MD, MBA, said upon introducing the program in June. “Combining the speed of technology and the experienced clinicians, this new model helps bring Medicare into the 21st century.”
- AI expert to the Emergency Room—STAT. Patients and loved ones are unlikely to ever hear such an announcement over a hospital P.A. But they may as well, and soon, if present trends proceed apace. “In emergency medicine, AI has gained traction not only in clinical decision support but also in digital twin modeling of patients, predictive analytics for emergency department flow, and integration with prehospital emergency medical services (EMS),” explains PhD candidate Hugo Francisco de Souza in News-Medical.net. “AI represents a transformative force in emergency medicine with the potential to accelerate and improve the accuracy of patient triage, diagnoses and resource management, thereby leading to a more efficient and resilient global emergency care system.”
- What does hospital architecture have to do with healthcare AI? Maybe not much now, but things are changing. “The real opportunity of AI isn’t just efficiency—it’s creating space for empathy,” says a senior architect with the 70-year-old firm HOK Architecture. “That’s what designing healthcare experiences should be about: making human connection possible.” Of course he would say that. There are prospects to pitch and business to win. Still, architecture is an interesting planet to add to the growing healthcare AI universe. Hear out the HOK forward-looker at Building Design + Construction.
- Instead of waiting for illness to occur, AI could continuously monitor individual health, identifying risks before symptoms emerge. When a prediction like that comes out of monitor-mad China, people the world over listen—with some admixture of anticipation and trepidation. And China is exactly where this one comes from. Such a shift, Professor Xu Chuan from Chongqing Jinfeng Lab tells the Chinese news & information outlet CGTN, “would represent a fundamental rethinking of healthcare: an interconnected system that links homes, communities and hospitals into a seamless network. … For patients globally, it holds the potential to make healthcare more predictive, personalized and affordable.” That’s a sunny forecast. We’ll have to wait and see if any dark clouds roll in from the East.
- From AIin.Healthcare’s sibling outlets:
|
| | | |
| |
| | | Breakthrough: clinical audio hits 93% accuracy Ambient AI fails when accuracy stumbles. In testing, Speechmatics’ new medical model hit 93% transcription accuracy and logged 17% fewer errors than the next best provider. It separates speakers in real time, copes with accents and interruptions, and captures drug names, ICD-10-CM and procedure codes, plus timings. Run it live or pre-recorded audio. Read more.
The ambient AI playbook: Lessons from two leading health systems At the recent CompassionIT Summit, leaders from Akron Children’s Hospital and Denver Health shared powerful lessons from rolling out ambient documentation to over 1,500 clinicians. Their biggest takeaway? Stories, not stats, drive adoption. Whether it was a heartfelt testimonial that swayed an entire department or a 60-second Nabla demo that eliminated training anxiety, the common thread was simplicity, authenticity, and clinician-centered design. Read more about the way these health systems are navigating ambient AI implementation: https://dhinsights.org/news/the-ambient-ai-playbook-lessons-from-two-leading-health-systems
|
| | | |
|
| | | Conventional wisdom has it that older adults willingly lag behind younger subpopulations when it comes to reaping the benefits of emerging healthcare technologies, including AI. A new literature review shows this may be a tone-deaf diagnosis. What’s actually lacking, the review’s authors surmise from their findings, is research aimed at understanding the community healthcare needs of particular seniors. “Our systematic analysis identified a robust body of research on AI for older adults. However, a critical gap emerged with a dearth of studies explicitly focusing on older adults in rural communities,” write Kristina Shiroma, PhD, and Jacqueline Miller, MLIS, in a paper published Sept. 15 in JMIR Human Factors. “This lack of representation raises concerns about the generalizability of findings and the potential for exacerbating existing healthcare disparities in rural areas.” The researchers, both with the School of Information Studies at Louisiana State University, identify three themes that emerged from their study, which reviewed some 23 published studies. 1. Numbers over narratives.AI for health research predominantly emphasizes quantitative findings, focusing on measurable results, such as diagnostic accuracy, predictive modeling and algorithmic efficacy. “This trend reflects broader scientific priorities placed on numerical data and technical performance, often at the expense of deeper, qualitative insights into the lived experiences of older adults,” Shiroma and Miller remark. ‘Without [a] qualitative dimension, the development of AI technologies will likely continue to miss crucial elements that could optimize their utility in everyday healthcare settings, particularly in rural and underserved populations.’
2. Efficacy over impact.A more inclusive, participatory approach may ensure that AI tools are better aligned with the challenges and needs of rural older adults. “[W]hile AI-driven tools demonstrate potential in improving diagnostic precision and personalizing care for older adults, the understanding of their real-world application in rural areas remains underexplored,” the authors note. ‘[W]ithout a community-centered approach to AI for health research—one that includes understanding the barriers and experiences of rural older adults—the focus may remain on clinical outcomes rather than the broader community health outcomes that truly affect these populations.’
3. Deepening disparities.The findings from this systematic review of the literature highlight a glaring gap in the representation in AI for health research of the perspectives of older adults from rural areas. “AI tools for healthcare and health decision support are developed using large datasets that, if not inclusive, will continue to marginalize communities like rural older adults,” the authors comment. “Moreover, most of the studies we reviewed were heavily urban-centric, emphasizing data from metropolitan populations while overlooking the distinct challenges and health contexts of rural communities.” ‘AI models developed in these contexts may fail to reflect the lived experiences or healthcare needs of rural older adults, further widening existing disparities.’
Shiroma and Miller suggest future research into rural older adults vis-à-vis healthcare AI should: - Prioritize targeted recruitment strategies for rural older adult participants to ensure better representation in AI for health research;
- Develop community-based AI policies, practices and products that reflect the specific needs and contexts of rural populations; and
- Explore solutions that address the limited representation of rural communities, ensuring that AI interventions are equitable, accessible and beneficial for all.
The study is posted in full for free. - Other newsworthy research:
|
| | | |
|
| |
|
|
