Note: Views expressed by Guest Bloggers do not necessarily reflect the views of “It Takes a Community” or Allscripts.

While Meaningful Use-certified Electronic Health Records have become the de-facto standard and a must-have technology for most hospitals and increasingly clinics and physician practices, physician adoption of EHRs has been lagging – for well-understood reasons. Physicians have a strong need and desire to maximize their time with patients. Yet, technology is competing more and more for their time, and creating structured clinical documentation using EHR templates may actually negatively impact physician productivity in a point-and-click environment.

Despite the availability of significant amounts of federal stimulus dollars to Meaningful Use-compliant institutions, most physicians perceive this to be less critical to widespread EHR adoption than access to tools that simplify, speed-up and improve the quality of clinical documentation within their EHR. An essential part of improving the quality of clinical notes is the capture of a more complete patient story via physician narrative to complement the less expressive point-and-click templates. Without such narrative, and due to widespread copy-and-paste-forward behavior, clinical notes lack documentation of the physician thought process that is so critical in coordinating care among multiple care providers. In the words of one physician: “I am looking at eight pages of detailed, structured EHR documentation and still have no clue what’s going on with this patient.”

This explains why physician narrative has always been the most popular mode of physician documentation. On average, 60 percent of a patient’s medical history still resides in unstructured clinical documentation. And with the recent availability of powerful speech and natural language understanding tools that liberate the clinical content hidden inside those narratives, that number is likely to go up rather than down. In the past, narrative clinical documentation was simply neither accessible nor actionable from a computational perspective – but that has changed rapidly in the last few years.

Today, we can capture validated clinical data directly from physician narrative in real-time, and use it to drive Meaningful Use and Quality Measures reporting, abstracting and billing needs – all of that without making physicians point-and-click through pages and pages of structured templates. This not only delights physicians by allowing them to spend more time with their patients again, it also results in more meaningful clinical documentation that physicians find more useful as it tells the patient’s story and contains all the relevant clinical facts and considerations, rather than putting too much emphasis on just the reimbursement requirements.

The recent advances in Speech and Natural Language Understanding technology are giving physicians the best of both worlds – efficient narrative documentation and compliance with regulatory and financial requirements – something that traditional speech recognition technology is not capable of delivering. These new technologies implement a closed loop documentation workflow where physicians receive guidance about documentation requirements and compliance based on an understanding of the content of their narrative documentation. This allows the physicians to get to the right level and specificity of clinical documentation (key use case: documenting appropriately for ICD-10) in a once-and-done workflow, as opposed to receiving queries from coders or CDI specialists at a later time when it is a huge disruption.

QUESTION: With the continuous increase in regulatory requirements relative to clinical documentation (Meaningful Use, quality measures, ICD-10, value-based purchasing), healthcare organizations have to ask themselves: how do we keep physician productivity and quality of clinical documentation high while driving adoption of our EHR? If you have a minute, let me know your thoughts on this!

You can hear more on this topic at ACE 2012, where I will present on “Driving Adoption of Clinical Noting Through Speech Understanding.” August 16 at 11:45am in W196A.




Tags: ,

About the author

Dr. Juergen Fritsch serves as M*Modal’s Chief Scientist. Dr. Fritsch co-founded MultiModal Technologies, Inc. in 2001. Prior to MultiModal Technologies, Inc., Dr. Fritsch was one of the founders of Interactive Systems, Inc., where he served as Principal Research Scientist and was responsible for developing a next generation medical speech understanding system. Dr. Fritsch held research positions at the University of Karlsruhe and at Carnegie Mellon University. He earned his M.Sc. and Ph.D. degrees in computer science from the University of Karlsruhe.

2 COMMENTS on Using Speech Understanding™ to Power Clinical Documentation, Technology Adoption


Nate Evans says:

07/18/2012 at 9:48 pm

I’m a fan of technology and have often wondered about speech recognition as a tool to assist physicians. I also wonder, how may this component impact errors in the record keeping process. Is there a net improvement, or a system in place to ensure the correct medications are transcribed as the physician mutters complicated medical terms?

Also, it’d help if you altered your product to read research articles out loud to me, maybe by cooperating with good reader or some other document app. That’d be useful. Think of the positive marketing benefit. Using this medically accurate reading software may support student health professionals to adopt your technology once they finish school and get to work.

I’d pay for that, if there were student loans left over… But I’d much prefer it for free even if program kept telling me about Allscripts etc, as I opened it.

Juergen Fritsch says:

07/23/2012 at 8:56 am

Dear Nate, you raise an interesting and valid question: how does speech-based noting affect overall clinical documentation quality? The technology certainly isn’t perfect and may introduce errors due to varying speech patterns and other factors. But the latest speech understanding technology is actually capable of annotating the transcribed text with the likelihood of there being an error. This information is sometimes presented to the user in the form of color-coded text that draws the attention to the areas that are most likely in need of correction. Irrespective of that though, users always have to carefully proofread their dictations for such errors. With that in place, the fact that speech-based reporting enables a rich narrative about the patient’s unique story has been shown to lead to higher quality documentation compared to point-and-click only templates, striking a balance between the use of speech and template-driven environment. Such documentation contains more detailed and more meaningful information about the physician’s thought process and that enables the next physician who sees a patient to make better decisions by having access to information that is typically not captured in structured templates.


Your email address will not be published. Required fields are marked *