CHICAGO—3M Health Information Systems has tapped Amazon Web Services to use machine learning and generative AI technology to advance automated medical notetaking and virtual assistant solutions for doctors.
The company plans to use Amazon Bedrock, Amazon Comprehend Medical and Amazon Transcribe Medical to scale its AI-enabled ambient clinical documentation solutions and speech recognition tech with the aim of reducing paperwork burdens for physicians.
It's the second major partnership that AWS announced this week at the Healthcare Information Management and Systems Society Global Conference. AWS also is collaborating with Philips to send the imaging tech company's storage and sharing platform for imaging scans to the cloud and to bolster that platform with new artificial intelligence-powered capabilities.
Between a range of collaborations across healthcare, pharma and medtech, Amazon has spent the last few years moving deeper into the world of medicine—and these newly announced partnership are strengthening its foothold in the industry.
Working with AWS, 3M HIS will advance its conversational AI platform, which is currently used by more than 300,000 clinicians. The company offers cloud-based solutions such as 3M M*Modal Fluency Direct for real-time speech recognition compatible with more than 250 electronic health records (EHRs) and 3M M*Modal Fluency Align for ambient clinical documentation.
The global industrial and technology giant bought healthcare speech recognition company M*Modal back in 2018 for $1 billion.
Working with AWS, 3M will build out ML-based clinical documentation and virtual assistant solutions that integrate directly into workflows and help ensure that the physician is in control of the information being entered into a patient's health record, executives said.
AWS is focused on providing the "building blocks" to enable its healthcare customers to develop the next generation of cloud-based solutions, Rowland Illing, M.D., chief medical officer and director of international public sector health at Amazon Web Services (AWS), told Fierce Healthcare this week.
Last week, AWS announced its new Bedrock service to help build and scale generative AI applications, which are applications that can generate text, images, audio, and synthetic data in response to prompts. Foundation models (FMs) are the ultra-large ML models that generative AI relies on, according to Illing.
"This enables us to scale foundation models to third parties," Illing said. "The mission of what we are doing now is democratizing access to foundation models. With this announcement from 3M, they are leveraging our new Bedrock service to developed the next generation of clinical pathway management solutions, ambient voice and ambient listening solutions to clinical end users."
He added, "This is hugely exciting and it's a growing area and there's a lot of demand for it in the market. That's why we've been launching a number of different healthcare-specific services such as Amazon HealthLake, HealthLake Imaging, Amazon Omics for genomic data analysis. Bedrock is just another component of what we see as a number of services that will be daisy-chained together to build solutions in the future."
While there is currently growing excitement about generative AI, it represents just one of many different types of foundation models as computing power can now leverage massively large datasets. AWS wants to play a role to do the "heavy lifting" to make these technologies available to the industry, Illing said.
"There are a whole slew of AI companies who are releasing their foundation models available on the Amazon Bedrock service. What customers can do is they can leverage the foundation models, which are basically pre-trained models on massive datasets," he said.
In 2019, the largest large model was around 330 million parameters and now large language models are upwards of 500 billion parameters, he noted.
"It's really the power of cloud that's enabling all of this to happen. Without these virtually unlimited amounts of storage and compute, trying to do this kind of analysis on 500 billion parameters worth of data, people just don't have enough capacity for their on-premises computing," Illin said. "Cloud will be the way of the future. I think we're going to see all of these clinical systems infused with AI and ML in the future.
Pairing 3M cloud-based clinical intelligence with AWS ML services will further help enable 3M ambient clinical documentation solutions to unobtrusively support the complex task of documenting the patient interaction.
3M's collaboration with AWS expands on its early success in bringing conversational AI and ambient intelligence directly into clinical documentation workflows through 3M Fluency Align. Working with AWS will make it easier for clinicians to automate accurate, complete and structured notes in the EHR a scalable reality, executives said.
Using contextual understanding, 3M Fluency Align takes the patient-physician conversation and available EHR data to create a quality-reviewed note directly in the patient record, ready for physician review and sign off. Using Amazon Bedrock's generative AI service, 3M will further scale and accelerate its innovation in conversational AI with the goal of providing greater flexibility, choice and usability to physician end users, 3M executives said.
"We believe that 'language is the highest form of intelligence' and that the smart use of AI in health care can super-charge our clinical ability. 3M M*Modal is a partner we rely on. 3M's ambient capability allows our doctors to focus on what they do best --- 'the Medicine.' When built into clinical workflows, this allows us to focus on direct patient care and makes health care a richer experience," said Dr Shankar Sridharan, consultant paediatric cardiologist and CCIO, Great Ormond Street Hospital, U.K., in a statement.
Philips also will use AWS' newly launched Amazon Bedrock service to build generative AI tools, powered by machine learning algorithms dubbed Foundation Models that are designed to sort through massive amounts of data and perform more complex tasks than standard AI models, Fierce Medtech reported.
Those AI tools will be integrated into Philips' HealthSuite platform to help hospitals by automating administrative tasks, providing doctors with decision-making support and even guiding them toward more accurate diagnoses—thanks to machine learning technology that can sort through imaging data more quickly and incorporate voice recognition abilities into those analyses.
Illing sees AWS' role in healthcare to enable and democratize access to innovative technologies.
"We're going to keep on building those services. We're going to keep on asking our end customers what they want and health techs what they need in order to provide the right building blocks because that's the way that will have the biggest impact," he said.