listen to article
This voice is automatically generated. Please let us know if you have any feedback.
LAS VEGAS — Executives from America's largest technology companies gathered at this year's HLTH conference, pledging to lead the next innovation in medical artificial intelligence.
AI products from companies like G.Woogle, Microsoft, Amazon, GE Healthcare, Nvidia, We promise our systems will help you solve a variety of problems, from reducing documentation time to optimizing operating room schedules. Technology executives say they have moved beyond early AI tools, which tended to be single-point solutions, to offering platform solutions that can be tailored to the needs of individual health systems.
“We are truly at a tipping point,” said Sally Frank, global head of health and life sciences at Microsoft.
But as tech companies seek to sell their products to health systems and providers, they are unsure how to responsibly bring new technology to the industry and how clinicians are similarly prepared to respond to new tools. Executives are divided on whether or not there is.
“Any evaluation we do at the basic model level is just a first draft,” he said. Greg Corrado, senior director of Google Research and co-founder of the Google Brain team. “Health systems with the desire and capacity to conduct research on the ground need to be the first movers, and not all health systems can do that.”
Technology companies value platform approach
Technology companies market themselves as partners for healthcare organizations in AI development, offering deep expertise in product development and testing.
But HLTH technology executives cautioned that there is little in the way of telling health systems how to apply AI.
“Google is not a healthcare provider, and we don't want to be one,” Collard said.
The executive said AI use cases should come directly from health systems. Technology companies are like semiconductor chip makers, Corrado said, building valuable, resource-intensive raw materials — in this case, large language models — that the medical industry applies.
“The technology we develop must enable healthcare organizations to envision and build their own future in this space,” he said.
Google has been working with health systems like Cleveland Clinic and Community Health System to pilot technologies ranging from health-specific cloud service platforms to document tools for searching electronic health records.
Other technology companies, such as Microsoft, are developing similar partnerships that encourage health systems to customize their own AI solutions.
it's a promise Kees Hertogh, Microsoft's vice president of health and life sciences product marketing, says this opens the door beyond “out-of-the-box” tools that solve “simple use cases.” .
Earlier this month, Microsoft announced that the health system Build your own AI tools directly. The company also utilizes Generative AI, which allows health systems to organize unstructured data, image data, and medical imaging data, allows clients to “build their own co-pilots, their own AI agents,” Hertogh said..
Health systems are likely to prefer these tools over older products, said Bill Ferra, principal consultant at Deloitte.
“As people get more comfortable with doing their own builds and understand how to do it, I think hyperscalers and platform players will take over this space,” Fera said in an interview. “There will be a shift away from applications to proprietary platforms.”
The need for robust testing raises questions about access
In healthcare, tech companies say they are now interested in helping health systems manage information overload. The company's ability to organize large amounts of data in written format as well as images reduces the time it takes for healthcare providers to review notes or schedule patients for important surgeries. executives say it could help.
However, different AI applications require different levels of human oversight.
For now, HLTH technology companies agree that all healthcare AI requires a human supervisor. Some companies focus first on management-focused AI tools because they require less oversight. For example, at GE Healthcare, scheduling tools may require less administration than tools used to support cancer treatment.
Abu Mirza, general manager and global SVP of digital products at GE Healthcare, said: “We believe we can deploy AI very quickly in these areas because it ties directly into clinical decision support. ”. “It is not directly related to someone's decision about surgery or anything else. This is where AI really starts. ”
But as AI approaches clinical decision points, more robust testing will be needed, Google's Corrado said.
The problem of AI “hallucinations,” or models producing answers that are not supported by the source text, has received a lot of attention.
But Corrado argued that omissions should be treated with equal caution. Omissions occur when the AI fails to cite relevant information in its answer. In some ways, omissions are difficult to examine because they often require a complete review of lengthy medical records.
But Microsoft and Google executives say not all health systems have the resources or know-how to reliably test cutting-edge technology.
Hertogh said Microsoft has connected healthcare systems seeking to leverage AI, allowing more advanced systems to provide resources and knowledge to smaller systems before deployment. The consortium includes more than 15 hospitals and health systems. Providence, Advocate Health, Boston Children's Hospital, Cleveland Clinic, Mercy, and Mount Sinai Health System.
Microsoft provides voluntary testing guidelines for health systems using AI products.
“We provide them with guidance and technology so they can achieve visibility and transparency into their systems… That relationship is about, 'Hey, this is how we approach building; This is kind of how we think about responsibility' AI,'' Hartog said.
Microsoft also benefits from the information exchange, the executive said. Microsoft has a view on how AI is implemented, but this interaction is a two-way street. Engaging with leaders of the nation's best health systems will help foster trust and establish credibility in Microsoft's services, he said.
According to Hertogh, this widespread spirit of cooperation is relatively new. He believes it could help small healthcare providers access new technology and act as an “accelerator” for broader adoption of generative AI across the healthcare system.
But Google doesn't just suggest that health systems test AI; the company is almost requiring it.
Collado said clinical feedback is integral to Google's testing process, so so far the company has only worked with health systems that have some “advanced technology” when it comes to AI and share its values around model testing. He said he is affiliated with the company.
“I think healthcare organizations that want to buy an off-the-shelf solution should probably wait. Until things mature and stabilize,” the researchers said. “I don't think there's a mature enough assessment framework out there that you can put technology into right away in any[health care]setting. … There's nothing as ironclad as this. We need to evaluate how it will actually be used in the field and make adjustments accordingly.”
