The problem: Intense Clinician EHR Interactions are Compromising Patient Care
During a panel presentation at a 2018 AAMC meeting in Washington, DC, I was asked what I thought was the one thing that could be done to improve the usability of and satisfaction with the EHR.
I remembered reviewing myriad physician comments from the KLAS Arch Collaborative research and the phrase that had stuck with me the most: “The EHR is causing my death by a thousand clicks.” So of course, my suggestion to the AAMC meetings’ attendees was simply, “Ditch the keyboards!” My response shocked the audience, and their shock shocked me.
We frequently hear the frustration about “too many clicks” in the EHR. One way to reduce clicks is to implement voice recognition solutions. I am one of the baby boomers who grew up watching Captain Kirk or Spock simply address the computer with spoken commands. It appears that we may be getting closer to this reality.
The Solution: Leveraging Voice Recognition Technology to Improve Care Delivery
Front-end voice recognition solutions that capture physician speech and convert the speech into patient documentation or notes have begun surfacing in ambulatory clinics. This has created an emerging market of “virtual scribes.”
These virtual scribe solutions expand traditional data-entry capabilities with new technologies by:
- Using an “Alexa” or Google Glass-style device in the exam rooms
- Using NLP to extract data from care visits to support billing and referrals
- Deploying APIs for more efficient interoperability with EMRs.
Providers should exercise caution and evaluate these new offerings on their security capabilities. All documents and data created from these solutions should be encrypted at rest and in transit. Additionally, some of these solutions still require human intervention in the creation and review processes, as some studies have reported an error rate above 7 percent in notes generated by virtual scribes.
While some organizations have leveraged speech recognition among nurses for years, I suspect that the low adoption rates have been due to the mobile nature of nursing. Integrating NLP into a mobile device for nursing would likely drive higher adoption rates.
Inching healthcare ever closer to Star Trek, some organizations now use voice recognition for EHR navigation. OrthoVirginia, for example, implemented a tool from M*Modal, enabling physicians to conversationally interact with Epic’s mobile app in order to look up patient information, lab results, problem and medication lists, and visit summaries. Imagine the possibilities!
The Justification: Moving the Needle on Burnout
While most voice recognition and NLP solutions have focused on billing support in order to drive the ROI for their solutions, the real value of this technology lies in quickly extracting clinical data for care visits. Seamless data entry would better facilitate diagnostic guidance, clinical decision support, and analytics via interoperability with the EHR.
Beyond that, the ability to navigate the EHR functions with voice recognition instead of keyboards frees clinicians from the workstation. This change ultimately allows them to spend more time with the patient and less time pecking away at the keys. After all, as Star Trek’s McCoy might say, “They’re doctors, not data-entry clerks!”
Players in the Market Today
Example companies and solutions include Suki, Robin Health, Augmedix, Saykara, and Notable. Microsoft (Empower MD project), Amazon (Comprehend Medical), M*Modal, and Nuance are examples of large, established companies who are entering this field that will help drive advancement and adoption of these solutions.
- Pay particular attention to the vendors who are demonstrating success across a wide array of clinical specialties with their NLP capabilities. This capability depicts the strength of the NLP ontology, and will likely enable easier pathways for extending these solutions to other services.
- Be aware of proprietary technologies being used for these solutions. In many cases, proprietary solutions are needed to deliver the necessary capabilities, but in the long run, proprietary technologies can be expensive to maintain and support.
The voice recognition and NLP market is emerging and therefore represents a high level of risk as provider organizations work to achieve successful implementations and get ongoing support. Providers should evaluate voice recognition and NLP solutions in small and controlled specialty environments in which the vendors in question have demonstrated success.
Once a tool has been proven in those areas and the organization has achieved proficiency with a product, the organization can then expand to other environments.