News

Bringing Voice Services to the Patient Bedside


Posted in Mobile Health by mthibault on April 18, 2017

A digital health innovator discusses how voice platforms are being used in the hospital and home today and what healthcare services may soon be possible using the technology.


Marie Thibault

Visitors to Boston Children’s Hospital might notice a familiar accessory on some ICU charge nurse stations and mobile computer carts: Alexa-enabled Amazon devices. The institution has pioneered work with the voice service, including developing the first health skill on the device.  

The KidsMD Alexa skill was created to give parents a resource and decision-support tool for questions related to their child’s fever or other symptoms. Matt Murphy, innovation lead at Boston Children’s Hospital Innovation and Digital Health Accelerator, described the effort as a collaboration between technologists—software developers and machine learning experts—and clinicians.

“Parents ask a question, it calls our system, and our system responds back with data and symptom information that is read off to the parent,” Murphy explained.

KidsMD is just the start of the hospital’s interest in voice platforms. Murphy’s team is developing skills with the Alexa voice service for use inside the hospital, too. “The nice thing is that it’s such a flexible platform that we’re doing a lot of work independently within the enterprise,” Murphy said. “Alexa, for us, is more about leveraging this initial platform as a way to experiment with voice more broadly in the hospital setting.”

Learn more from Matt Murphy during his session, "Alexa Will See You Now: IoT Technology for the Next Generation" at the BIOMEDevice Boston Conference and Expo, May 3-4.

Boston Children’s is including a variety of stakeholders in the process of developing these applications. Patients, parents, social works, clinicians, and others were part of a hackathon in late 2016 to develop voice service skills for the hospital environment. One such skill was intended to assist parents who care for children with peripherally inserted central catheter (PICC) lines, instructing them in each step of cleaning the PICC line, changing medications, and addressing simple problems at home.

Patient-specific information is not available via voice command because voice platforms are not yet HIPAA compliant. Still, there may be ways to use voice services to give users information to schedule or confirm appointments, Murphy pointed out. “We definitely need to make that transition into the HIPAA-compliant framework before we can build out fully-enabled skills that have all the patient’s health information on demand.”

Still, there are plenty of useful skills that are possible while steering clear of patient-specific information, such as enabling a user to pinpoint the location of a specific device or medication within a vast stock room. Murphy added that the team also developed voice skills for the Intensive Care Unit (ICU), giving nurses support regarding lab information, blood draws, and related uses. That led to voice commands being used in the ICU today to offer basic reference information on medication dosing, specific protocols, contact phone numbers for staff members, and other useful information.

While exploring the use of voice platforms for healthcare, Murphy’s team has picked up useful lessons along the way. For example, hospitals can be noisy, making it tough for the device to identify commands and for the user to hear a response. “We did have to do a lot of work around optimizing placement of some of these devices to capture what a clinician is saying in a real setting,” Murphy said. That involved experimenting with various locations for the device and learning where a clinician is in a room when they’re using a voice skill.

The team found that one of the best ways to gain support for the voice skills from clinician users was to use the platform for easier, quicker access to reference information, Murphy said. Instead of shuffling through massive binders and piles of papers or performing numerous online searches for necessary information, doctors could ask the Amazon Echo device for the answer. “Prioritizing those as the first skills to develop was really powerful and got a lot of buy in from clinicians who were engaging with the platform,” Murphy added.

Another lesson? The delivery method for information—verbal versus visual/reading—changes the way information is best presented. The information being referenced by the voice platform had to be reorganized to prioritize information for the user. “Voice is a great interface,” Murphy said, “but it’s not one where you want to be reading off paragraphs and paragraphs of information.”

To address this, Murphy’s team sought the advice of user experience designers to rethink the user experience with voice skills. “Thinking of prototyping via conversation versus prototyping via wire frames like you would do for web has been a great way for us to really create skills that are useful and lightweight to use on demand,” he said. The verbal delivery also meant the team had to hard code specific pronunciations for certain medical terms and medications.

The efforts at Boston Children’s Hospital offer a glimpse of the incredible potential for voice applications in healthcare. “I’d say there’s definitely a future vision of voice being able to drive a lot of the interface or interactions that clinicians have with the medical record and other reference data sets around patient information, just because it’s such a deep amount of information you have to process about a patient,” Murphy said. “Pulling those things via voice commands is really powerful and I think that’s probably the future of voice in healthcare.”

Marie Thibault is the managing editor at MD+DI. Reach her at marie.thibault@ubm.com and on Twitter @MedTechMarie.

[Image courtesy of PONG/FREEDIGITALPHOTOS.NET]


Tags:
Printer-friendly version
Your rating: None Average: 4 (1 vote)

Login to post comments