Wall Street Journal
By Isabelle Bousquette
March 24, 2022
Mayo Clinic in Rochester, Minn. The Mayo health system also has campuses in Arizona, Florida and several other locations.PHOTO: KEREM YUCEL/AGENCE FRANCE-PRESSE/GETTY IMAGES
Key messages
by Joaquim Cardoso MSc.
ai and digital health . institute
March, 27 2022
What is the problem?
- Some 80% of patient data in electronic medical records is unstructured, including transcribed reports and doctor’s notes
- We just have a lot of data that is not accessible and locked down because they’re in an unstructured format
What is the solution?
- Natural language processing allows the hospital system to turn the data that isn’t organized into discrete fields into structured data, meaning clinicians will have an easier time searching and analyzing it, he said.
- Once patient data is structured, the hospital will be able to find ideal patients for clinical trials and run various types of prediction tools that look for indicators that certain patients might be at higher risk for certain diseases early on, said Dr. Anantraman.
- This type of work was possible in the past, but required a lot of human involvement, he said.
- This technology, known as natural language processing, has made strides in recent years, say experts, although it remains early days.
Who is doing it?
Mayo Clinic and Google
- Mayo Clinic said it is assessing a product from Google Cloud that uses artificial intelligence to better track and analyze language-based information in patients’ electronic health records, such as doctor’s notes.
- Mayo Clinic has used natural language processing tools to index unstructured information from medical records in the past, but Dr. Anantraman said the Google technology is more scalable and would allow Mayo to do so more accurately and efficiently.
- Google’s Healthcare Natural Language API could help clinicians search and access patient data a lot more easily, although he said Mayo is still in early trial phases.
Alphabet
- Alphabet Inc.’s cloud unit is one of many technology companies pouring investment and research into the area.
Microsoft Azure
- Microsoft Azure’s Text Analytics for Health and Amazon Web Service’s Amazon Comprehend Medical also continue to roll out new capabilities for their natural language processing for healthcare offerings.
- Microsoft Corp. completed its acquisition of AI speech-recognition company Nuance Communications Inc. earlier this month.
- Providence and Microsoft have been partnering since July 2019 and have built out cancer-specific natural language capabilities that use Microsoft’s Text Analytics for Health.
What other technologies are under development? (Voice Technology)
- At Houston Methodist Hospital, Chief Innovation Officer Roberta Schwartz, said she has been working with Amazon on voice technology that: transcribes physicians’ conversations with patients, translates that conversation into notes, and then translates those notes into discrete fields in the medical record.
ORIGINAL PUBLICATION (full version)
Mayo Clinic Assessing Google AI Tool as It Seeks More Insights From Patient Records
Wall Street Journal
By Isabelle Bousquette
March 24, 2022
Mayo Clinic in Rochester, Minn. The Mayo health system also has campuses in Arizona, Florida and several other locations.PHOTO: KEREM YUCEL/AGENCE FRANCE-PRESSE/GETTY IMAGES
Mayo Clinic said it is assessing a product from Google Cloud that uses artificial intelligence to better track and analyze language-based information in patients’ electronic health records, such as doctor’s notes.
This technology, known as natural language processing, has made strides in recent years, say experts, although it remains early days.
Alphabet Inc.’s cloud unit is one of many technology companies pouring investment and research into the area.
Vish Anantraman, chief technology officer of Mayo Clinic, said Google’s Healthcare Natural Language API could help clinicians search and access patient data a lot more easily, although he said Mayo is still in early trial phases.
Google’s Healthcare Natural Language API could help clinicians search and access patient data a lot more easily, although he said Mayo is still in early trial phases.
Some 80% of patient data in electronic medical records is unstructured, including transcribed reports and doctor’s notes,
Some 80% of patient data in electronic medical records is unstructured, including transcribed reports and doctor’s notes, Dr. Anantraman said.
Natural language processing allows the hospital system to turn the data that isn’t organized into discrete fields into structured data, meaning clinicians will have an easier time searching and analyzing it, he said.
Natural language processing allows the hospital system to turn the data that isn’t organized into discrete fields into structured data, meaning clinicians will have an easier time searching and analyzing it, he said.
This type of work was possible in the past, but required a lot of human involvement, he said.
Once patient data is structured, the hospital will be able to find ideal patients for clinical trials and run various types of prediction tools that look for indicators that certain patients might be at higher risk for certain diseases early on, said Dr. Anantraman.
This type of work was possible in the past, but required a lot of human involvement, he said.
Mayo Clinic has used natural language processing tools to index unstructured information from medical records in the past, but Dr. Anantraman said the Google technology is more scalable and would allow Mayo to do so more accurately and efficiently.
the Google technology is more scalable and would allow Mayo to do so more accurately and efficiently.
“We just have a lot of data that is not accessible and locked down because they’re in an unstructured format,” said Dr. Anantraman, adding that the data is now much more accessible.
The application of AI to healthcare has met with varying degrees of success, complicated by factors such as access to patient data sets reflective of the population, lack of healthcare expertise on the part of tech vendors and hype.
The application of AI to healthcare has met with varying degrees of success, complicated by factors such as access to patient data sets reflective of the population, lack of healthcare expertise on the part of tech vendors and hype.
International Business Machine Corp.’s decision to sell assets from its Watson Health business, which heralded itself as a way to help doctors diagnose and cure cancer, highlighted the challenges various AI healthcare tools have faced trying to make an impact.
Still, the trillion-dollar healthcare industry remains a key target.
Tech companies point to ongoing investments in AI products, advances in the technology and natural language processing in particular, as well as partnerships with hospitals to collaborate on tools.
“We think [natural language processing] is the next frontier in healthcare,” said Aashima Gupta, director of Healthcare Solutions at Google Cloud, adding that the pandemic has increased the appetite among healthcare providers for the technology.
“We think [natural language processing] is the next frontier in healthcare,” said Aashima Gupta, director of Healthcare Solutions at Google Cloud, adding that the pandemic has increased the appetite among healthcare providers for the technology.
Stephen Messer, co-founder and vice chairman of Collective[i], an AI company focused on optimizing the sales process, said natural language processing technology has gone from “laughable to jaw-dropping in the span of two years.”
He said that is mostly because AI began training itself on larger and larger data sets.
Microsoft Azure’s Text Analytics for Health and Amazon Web Service’s Amazon Comprehend Medical also continue to roll out new capabilities for their natural language processing for healthcare offerings. ]
Microsoft Corp. completed its acquisition of AI speech-recognition company Nuance Communications Inc. earlier this month.
The Journal earlier reported that Nuance could boost its offerings in healthcare.
Microsoft Azure’s Text Analytics for Health and Amazon Web Service’s Amazon Comprehend Medical also continue to roll out new capabilities for their natural language processing for healthcare offerings.
Andrew Ng, an AI expert and CEO of startup Landing AI, said he is seeing more AI pilots in healthcare that are paying off.
“There are exciting examples where this work can make a difference, but these ideas will take years to diffuse,” he said. “I think we’re still many years away from being mainstream.”
Andrew Ng, an AI expert and CEO of startup Landing AI, said he is seeing more AI pilots in healthcare that are paying off.
At Houston Methodist Hospital, Chief Innovation Officer Roberta Schwartz, said she has been working with Amazon on voice technology that
- transcribes physicians’ conversations with patients,
- translates that conversation into notes,
- and then translates those notes into discrete fields in the medical record.
At Houston Methodist Hospital, Chief Innovation Officer Roberta Schwartz, said she has been working with Amazon on voice technology …
“It’s getting there,” Dr. Schwartz said. “We’re not throwing in the towel because we do think this is the future.”
Carlo Bifulco, medical director of genomics at Providence, said he is optimistic about the possibilities of this tech to affect patient care.
Providence and Microsoft have been partnering since July 2019 and have built out cancer-specific natural language capabilities that use Microsoft’s Text Analytics for Health.
“Things are continuously improving,” he said.
“I think the results are very promising,” Mayo Clinic’s Dr. Anantraman said of the Google trial. “I would refrain from saying it’s spectacular.”
Originally published at https://www.wsj.com on March 24, 2022.
TAGS: AI Health; Big Techs; Data Driven Health Care; Unstructured Data; AI Technologies; NLP; Data Accessibility, Future of Health Care, Future of AI
Names mentioned
Vish Anantraman, chief technology officer of Mayo Clinic
Aashima Gupta, director of Healthcare Solutions at Google Cloud,
Stephen Messer, co-founder and vice chairman of Collective
Andrew Ng, an AI expert and CEO of startup Landing AI,
Carlo Bifulco, medical director of genomics at Providence
