listen to article
This voice is automatically generated. Please let us know if you have any feedback.
Diving overview:
google cloud Expanding the suite of AI tools for healthcare professionals On Thursday, we launched a generative AI search tool for payers and providers to scrutinize. Create patient notes and scan documents and other data for clinical information.
The tool, Vertex AI Search for Healthcare, was first released in limited availability in March. The tech giant also expanded its healthcare data engine. Recording of patient data in industry standard formats.
Google promises that the risk of the tool outputting AI “hallucinations”, or inaccurate responses, is minimal.
Dive Insight:
A study conducted by Google Cloud and Harris Poll found that clinicians and claims staff log 27 and 36 hours of data per week, respectively. Administrative tasks including documentation.
Google says Vertex AI Search aims to reduce that time by allowing employees to query different aspects of a patient's medical record. This tool integrates with Search for Gemini 1.5 Flash and MedLM (two of Google's large language models).
Vertex was first teased by Hims March conference. Used by health systems such as Community Health Systems and Highmark Health.
According to the release, Google also requires Vertex AI to cite sources and provide links to internal sources (a technique called “grounding”) to increase provider confidence in the tool's answers. It is said that it promises.
This safeguard was introduced amid growing concerns that generative AI could output false or misleading information.
A study conducted in August by researchers at the University of Massachusetts Amherst and scientists at clinical AI company Mendel found that: Generative AI hallucinated when performing medical queries. AI tools created from research Medical record summaries contain false or misleading statements that can cause health care providers to misdiagnose patients or recommend inappropriate treatment plans.
Google's use of grounding is a “meaningful step” towards addressing hallucinations, it said. Jeffrey Cribbs is a renowned vice president analyst on the healthcare team at consulting firm Gartner. However, specifying the source of information can make the tool less effective in its intended purpose of reducing administrative burden.
“What I've just given you is a big research question,” Cribbs said. “The whole value proposition here is that we want to be able to save people's time. We can't expect them to go back to every part of their medical record. Otherwise, we're going to save them time. Therefore, over time, we want citations to be used as the exception rather than the rule.”
Cribbs said the industry will want to adopt a “higher level of validation” for generative AI so that providers can trust that the AI has correctly retrieved the provider's queries without manual review. .
Google has previously launched several generative AI ventures, including a large-scale language model tailored to search patient health records.
Cribbs said the company has become a leader in generative AI testing. Google is known in the industry for robust validity testing and checks on health equity measures amid concerns about bias embedded in its algorithms.
Still, Cribbs said validation and evaluation of generative AI is still an “emerging practice” and that some would-be consumers have concerns about search tools like the one Google announced Thursday. Of particular concern is that clinicians may come to trust AI output more than their own clinical expertise. This is a phenomenon called “automation bias.”
“There is very widespread concern among healthcare organizations considering the introduction of AI into clinical workflows about how the use of such tools could expose them to new types of risks,” Cribs said. he said.
Still, when implementing new AI tools, Cribbs emphasized that the goal is not perfection.
“We're not comparing it to the perfect information extraction that exists today,” Cribbs said. “The fact is that information is not being extracted from medical clinical records in an efficient or accurate way in today's market, so we are looking for significant improvements.”
