The Transformative Power of Conversational Technologies — Part 2
EHR optimization remains an elusive goal for many health care providers, but several digital assistants may help solve the problem.
By Paul Cerrato, senior research analyst and communications specialist, Mayo Clinic Platform and John Halamka, M.D., president, Mayo Clinic Platform
In our last blog we discussed the strengths and weaknesses of natural language processing (NLP), the engine that drives many conversational technologies. These innovations have given birth to several commercially available tools that are slowly helping improvement of EHR usability.* Few clinicians question the need for such improvements.
A 2020 paper from the National Academy of Medicine pointed out that 86% of office-based and 94% of hospital-based physicians used EHR systems and goes on to state: “While intended to improve care quality and efficiency, the EHR has inadvertently burdened clinicians and is now considered a leading cause of their burnout.” Solving this problem will undoubtably require a three-pronged approach that includes changes in policy, culture, and technology. Several vendors are attempting to address the third leg of the chair.
Nuance is one of several companies that have successfully incorporated voice technology into its products. The company is well known for its speech recognition software and has introduced what it refers to as ambient clinical intelligence (ACI) into its services. ACI allows clinicians to interact with their patients while the software does most of the heavy lifting; it listens to the conversation between physician and patient, automatically writing clinical documentation, including data concerning the history of the present illness, physical exam, assessment, and treatment plan. All the information is then input into the EHR. ACI takes advantage of voice biometrics, ML algorithms, and a 16-microphone–based platform. The company claims that when the patient and clinician speak naturally, it can translate non-clinical terms into clinical terminology and summarize natural language into coherent sentences.
Clinical decision support vendors are also taking advantage of advances in voice technology. UpToDate, for example, has incorporated Nuance’s Dragon Medical One into its system. Users can begin a clinical content search with words such as “Hey Dragon, search UpToDate for Type 1 diabetes.” If UpToDate users have an Anywhere enterprise license, it is possible to speak commands to obtain medication dosage schedules, drug interactions, clinical calculators, and much more.
M*Modal also has an NLP-enabled system that’s designed to create complete clinical reports in a clinician’s chosen EHR system. Fluency Direct uses a vendor-provided microphone and a cloud-based system that’s compatible with more than 250 EHRs. It features computer-assisted physician documentation (CAPD) functionality, which M*Modal says “continuously analyzes and monitors the clinical narrative. In real time, it nudges you for additional information or clarification and suggests specific things you can do to improve the quality of care and clinical documentation.”
Suki, another voice-driven digital assistant, also allows clinicians to speak their notes and orders into an EHR. It claims to reduce documentation time by 72% on average. It can copy notes from previous patient encounters, show the physician or nurse details on a specific patient’s medications, and retrieve their vital signs and problem lists. According to a report released by the American Academy of Family Physicians, “using an AI Assistant can significantly reduce documentation burden and family physician burnout.” The analysis looked at 132 clinicians, of which 60% fully participated. “These adopters saw a 72% reduction in their median documentation time per note. This resulted in a calculated time savings of 3.3 hours per week per clinician.”
While some clinicians may pine for the “good old days”, detailed electronic documentation of patient encounters has become an essential component of 21st century medicine. About 80% of the knowledge captured in an electronic health record comes from unstructured data; it’s critically important to have this narrative recorded accurately and comprehensively. Finding the right digital assistants will likely help achieve this goal.
*Footnote: Mention of commercial products does not imply endorsement by Mayo Clinic