Healthcare AI can only improve clinical outcomes if wins clinicians’ trust and does patients no harm.

The point is driven home by Daniel Yang, MD, vice president of AI and emerging technologies at eight-state, 12.5-million member Kaiser Permanente.

The need to succeed clinically is “why we use a responsible AI approach,” Yang writes in a piece posted this week. “With a focus on building trust, we use AI only when it advances our core mission of delivering high-quality, affordable healthcare services.”

Yang then summarizes seven principles that he says guide the integrated insurance/provider system—which employs 230,000 staff, 70,000 nurses and around 25,000 physicians—whenever it assesses AI tools for possible adoption.

These are:  

1. Privacy.

“AI tools require a vast amount of data,” Yang reminds before sharing:

Ongoing monitoring, quality control and safeguarding are necessary to protect the safety and privacy of our members and patients.

2. Reliability.

“What works today may not work a few years down the road as technology, care delivery and patient preferences evolve.” More: 

We choose AI tools that will work for the long term.

3. Eyes on outcomes.

If an AI tool doesn’t advance high-quality and affordable care, we don’t use it.

4. Transparency.

“We make patients aware of and ask for consent to our use of AI tools whenever appropriate.”  

For our employees who use AI, we provide explanations of how our AI tools were developed, how they work and what their limitations are.

5. Equity.

“People and algorithms alike can contribute to bias in AI tools,” Yang points out. “Our AI tools are built to minimize bias.”

We also know AI has the potential to harness large amounts of data and to help identify and address the root causes of health inequities, so we also focus on that potential.

6. Customer-centricity.

“In the case of AI, our customers are our members, doctors and employees who will use the tools.”

Tools must prioritize their needs and preferences.

7. Trust.

“We know there’s uncertainty about the effectiveness of AI,” Yang writes. “We choose tools that offer excellence in safety and performance, and alignment with industry standards and leading practices.”

We further build confidence by continually monitoring the tools we use. We continue to invest in research that rigorously evaluates the impact of AI in clinical settings.

Yang also encourages policymakers to help make sure healthcare AI gets developed and used responsibly. They can do this, he suggests, by:

• supporting the launch of large-scale clinical trials,  
   
• establishing systems to monitor AI tools used in clinical care and
   
• supporting independent quality assurance testing of AI algorithms.
   

Read the whole thing.

Carle Health chooses Nabla's ambient AI for it's focus on workflow - For a deeper look into Nabla's deployment at Carle Health, watch this insightful interview with Dr. David Lovinger and Ed Lee, MD, MPH, hosted by John Lynn.

Buzzworthy developments of the past few days.

• If you want still more from Daniel Yang, MD, you’re in luck. The Kaiser Permanente AI chief took a handful of quick questions from the Wall Street Journal. The newspaper posted the Q&A Wednesday. Yang says he’s been surprised by how quickly doctors have come around on AI since the generative type burst onto the scene. Many have shifted, he observes, from an attitude of reluctance and skepticism to one of “very deep excitement, delight and now demand.”
   
• Some 70% of healthcare consumers have no qualms about AI taking notes during doctor appointments. That’s according to a national sampling of 1,006 people. Ohio State University’s Wexner Medical Center commissioned the survey and further found 75% of respondents believing the use of AI to minimize human errors is “important.” The institution drew from the findings to guide a pilot program using the Microsoft Dragon Ambient eXperience (DAX) Copilot application. Project leaders report time savings of up to four minutes per patient visit. Some pilot participants have found the gain in efficiency translates to an upgrade in the human touch. “I’m spending as much if not more time with each patient,” says an internist, “and it’s higher quality time with more eye contact.”
   
• No one should be more excited about AI than nurses. Nurse.com makes the case without flatly stating it. The site lists 10 benefits—not least better patient care, increased efficiency and reduced workloads—and offers a quick overview of existing use cases. Acknowledging the worry among some nurses over AI’s potential to take their jobs, the author of the piece reassures readers: “[E]xperts say the unique qualities nurses bring to patient care, including compassion, communication, team collaboration and relationship building can’t be replaced by AI.”
   
• Enthusiasm over healthcare AI ran high in Verona, Wisc., this week. The small city (pop. 14,030) hosted tens of thousands from all 50 states and a bunch of countries when a star commercial resident, one Epic Systems Inc., convened a user’s group meeting. Company leaders told attendees Epic has more than 100 new AI features in the works for clinicians and patients. No less excitingly, CEO Judy Faulkner “stepped on stage to deliver a keynote dressed like a swan, feathers and all,” CNBC reports. Also worth a read is local coverage out of nearby Madison by the Cap Times.
   
• The NIH’s Office of Science Policy recently posted some essential new guidance. The document looks at how existing policies apply to healthcare research involving AI. Topics covered include participant protections, intellectual property, peer review and a good number of others. Lyric Jorgenson, PhD, the agency’s associate director for science policy, promotes the resource in a blog post. “As policymakers, we are continuously striving to develop policies capable of evolving alongside science and technology, taking into consideration that we know there will be unexpected twists and turns along the way,” she writes. “This is why we built a certain degree of flexibility into the NIH Data Management and Sharing Policy—because new tools for creating, sharing and accessing data are being developed every day.” Blog post here, guidance resource here.
   
• The Wall Street Journal tracked down a few Democrats at their national convention in Chicago this week. And peppered them with questions about federal AI legislation. For starters, what’s taking so long? “We’re going to get a great AI package which keeps innovation as our North Star, hopefully through the Congress by the end of the year,” Senate Majority Leader Chuck Schumer told WSJ reporters. “We have great prospects.” New York Assemblyman Alex Bores openly wished for federal regulation while expressing understanding about what’s been happening in its absence. “I don’t blame any state,” he said, “for jumping in front.”
   
• Healthcare AI can do a lot to improve quality of life for hurting patients. Take it from a teenager who relied on Siri and Alexa to help him through postsurgical challenges involving a bout with blindness. “The narrative of fear surrounding AI may be loud,” young Rishi Raja writes in the Los Angeles Times’s High School Insider, “but in my experience, its true power lies not in the darkness it may cast but in the light it can bring.”
   
• Dana Perrino has joined the American Health Information Management Association. No, not the Dana Perino who served as White House press secretary under George W. Bush and has attained arguably greater fame as a Fox News Channel personality. The Dana Perrino who’s now AHIMA’s chief member services officer spells her surname with two r’s, joined the org in 2021 and counts Meeting Professionals International among her former employers. Announcement.
   
• Recent research in the news:
   
  • Karolinska Institute: AI model aids early detection of autism
     
  • UCSF: Taming Parkinson’s disease with intelligent brain pacemakers
     
• AI funding news of note:
   
  • Caresyntax raises $180M to accelerate growth and adoption of precision surgery
     
  • Clarium announces $10.5M strategic financing round led by General Catalyst to accelerate the modernization of healthcare supply chain
     
  • Research funded by Curebound gains momentum with $4.1M follow-on funding for cancer research clinical trials and $1.2M for scientific prize recipients
     
  • California tried to make Google pay news outlets. The company cut a deal that includes funding AI
     
• From AIin.Healthcare’s news partners:
   
  • Health Imaging: AI rules out abnormal findings on chest X-rays, significantly reducing workloads
     
  • Cardiovascular Business: FDA approves new AI-powered imaging software from Nanox subsidiary