- Artificial intelligence is taking many forms in the healthcare environment as vendors strive to address deeply entrenched pain points for clinicians and patients.
Many electronic health record companies are integrating machine learning into their software offerings to streamline workflows, improve the patient-provider relationship, and hopefully prevent physician burnout.
But a handful of companies are considering an even more radical approach: removing the EHR as a presence in the exam room all together.
Google presented its vision for ambient computing at HIMSS18, garnering thunderous applause for the idea of “Dr. Liz,” an intelligent digital assistant that could create documentation, navigate the EHR, and present data to clinicians using solely the power of the human voice.
The company is pouring its resources into perfecting natural language processing (NLP) capabilities accurate enough for the clinical environment, and has published promising research on the topic.
Amazon has also dabbled in using its popular Alexa product as a virtual assistant for healthcare providers and patients, creating smart hospital rooms at Cedars-Sinai and creating an internal health and wellness team within its Alexa product development division.
Together, the Amazon and Google comprise almost all of the consumer market for smart speakers in the home, with Amazon enjoying a hefty 70 percent of the total market share.
While neither system currently boasts HIPAA compliance, it doesn’t seem like it will be long before the two technology giants capitalize on the opportunity to bring their ambient computing prowess into the clinic.
When they do, they will find that someone has beaten them to it.
At HIMSS19, held in Orlando, Florida in February, Nuance Communications showcased its Ambient Clinical Intelligence (ACI) solution, a tailor-made device that melds ambient detection tools with the popularity of its Dragon Medical virtual assistant to create a unique patient-provider experience.
The device, mounted to the wall, includes more than a dozen microphones that can capture and differentiate speech from up to eight individuals in the room, as well as a computer vision sensor that can track speakers as they move and ensure that speech is attributed to the right person.
A deep learning model then converts the speech into text, explained Peter Durlach, Senior Vice President for Healthcare Strategy, then translates the conversation into clinical documentation.
“The system can convert layperson terms into their clinical equivalents, so that providers can have natural conversation with patients and still capture what they need to support coding and clinical history,” he told HealthITAnalytics.com.
“So instead of requiring a physician to a conversation with a patient then type out the same information all over again, why can’t we extract the relevant information right from the start? Ambient clinical intelligence allows us to take the conversation, record it, and use deep neural net technology to extract the meaningful data and translate it into a structured document in the EHR.”
The goal of the technology is to address two of the most challenging ongoing issues in the healthcare environment: reducing physician burnout and creating meaningful, personalized patient experiences.
The two problems are intertwined. Even after close to a decade of widespread EHR use, providers are still incredibly dissatisfied with the need to spend hours each day in front of the keyboard, unable to devote their full attention to patients during shrinking appointment times.
The stress of endless clinical documentation and data entry is affecting the vast majority of organizations, recent research has shown, and extends past the physician to nurses and other healthcare professionals.
In 2018, a survey by the Physicians Foundation found that doctors strongly believe the EHR interferes with their ability to form meaningful relationships with their patients – a Stanford Medicine poll the same year also found that close to two-thirds of providers believe the EHR is a highly unwelcome presence in the consult room.
“Talking to any physician makes it very clear that they are frustrated with being unable to focus on the patient,” agreed Durlach. “But if you listen to a typical patient encounter, you know that much of what needs to end up in the documented clinical note is actually spoken between the participants.”
“Ambient technology allows providers to avoid so much of the administrative work and data entry that they dislike. They still get to confirm all of the orders and check over the documentation before anything is final – but if they don’t have to start from scratch after seeing their patients, they’re way ahead of the game.”
Nuance is also largely ahead of the game by creating a physical device that is healthcare-ready and HIPAA compliant.
“We don’t compete in the home with Amazon or Google, but we do power a number of virtual systems for brands like BMW, Ford, Delta, and American Express,” Durlach explained.
“We took all that expertise and technology, put a very strong HIPAA wrapper around it, and integrated it with our Dragon Medical flagship product. That gave us the ability to issue commands to the major EHRs, most of whom have already been our partners for decades.”
Nuance already works with Epic, MEDITECH, Cerner, and other companies to integrate voice recognition and transcription tools into the EHR workflow.
For example, Nuance voice recognition tools are available through Epic’s Haiku and Rover mobile applications, allowing users to retrieve patient information in a conversational manner. And 80 percent of Cerner customers also leverage Nuance tools for dictation and other documentation tasks.
“We don’t see this as taking anything away from the EHR, and we don’t view it as a replacement for the EHR itself,” Durlach asserted. “It’s an augmentation to a system that will remain vital for patient care.”
“There are so many things that EHRs do that we don’t do, like coding and clinical decision support. We aren’t an EHR, and we don’t want to be one. What we want is to make it easier for our customers to use the systems they have, and we’ve worked extensively with the EHR vendors to enable that.”
The company’s natural language processing tools are already well-known across the industry, particularly in the radiology space.
Durlach believes that familiarity will make adoption of an innovative approach to clinical documentation easier when the system becomes available in early 2020.
“We have the benefit of being a trusted partner in 90 percent of hospitals already,” he said. “We have the pillars already in place, so we’re pretty proud of leveraging that to get there first, before Google, Amazon, or any other competitor.”
“We’re looking forward to rolling out the ambient computing system and continuing on our journey to improve the experience of providers and patients across the industry.”