There is a growing demand to extract structured, “actionable” information from unstructured (dictated) medical documents. Clinical Language Understanding (CLU) technology allows a computer to read and understand electronic free text and extract data for use in countless applications across the healthcare spectrum. To understand and learn more about CLU technology, HIT Consultant spoke with Carina Edwards, VP, Solutions Marketing at Nuance Healthcare for a deep dive into:
- CLU Technology
- CLU technology implications for ICD-10 and Meaningful Use
- Nuance’s Partnership with 3M HIS
LISTEN BELOW OR CLICK HERE TO DOWNLOAD
HIT Consultant: Give me a brief overview of what exactly is clinical language Understanding (CLU).
Carina Edwards: So clinical language understanding (CLU) is the technology that Nuance has launched as a combination of natural language processing (NLP) and statistical analysis that allows us to take any form of documentation in text form, in speech form and extract the relevant clinical facts and codify them against the medical vocabulary. Be that, ICD-9, CBT, SNOMED, etc. It’s the technology itself that we refer to as the clinical language understanding (CLU).
HIT Consultant: What role does that play particularly for ICD-10?
Carina Edwards: So, it’s actually the utilization of clinical language understanding (CLU) that’s an important differentiator here. In it of itself, clinical language understanding is not a product perse’ and so what Nuance has done is we have embedded that technology into solutions that go to market with a specific use case. So, for ICD-10 as an example, we have two different solutions that are being brought into the market place. The first is called MD Assist. And we’ve worked with 3M to develop this solution and what it does it allows physicians, while they’re dictating, to look at the documentation, understand the level of specificity, and get prompted for more specificity when necessary.
So, let’s give a real life example. If I’m in the electronic health record and I’m using Dragon Medical to dictate into the electronic medical record. When I’ve done my dictation, and I have my full patient story documented, I’ll hit save, I’ll go to the next field for instance at that point in time, a query will prompt. So, if I had said, patient presents with heart failure. In the ICD-10 world, that would equate to about 50 or 60 codes, so I need more specificity. And so I prompt the physician to say, “What type of heart failure? What was the acuity? And the specificity of the heart failure?” So then quickly they have two radio buttons they can select: acute, systolic heart failure and now that goes right into the documentation and it’s one and done. So, that’s the MD Assist solution. And that’s on the front end capture. If you have the most specific documents, then the coders won’t have to go back, the clinical document improvement specialist won’t have to go back and query the physician to get further information to drive to appropriate reimbursement. Now that’s the front end solution.
On the back end, if you think about the full workflow of that document, we’re also putting clinical language understanding (CLU) inside of 3M’s 360 Encompass Computer Assisted Coding Application. As the document flows through the system, it goes into the 360 Encompass workflow, when the coder is presented with those facts, they’ll be extracted from the document using clinical language understanding and tagged to the ICD-10 code structure. Almost at that first pass from which then they can edit. And so, it’s a nice seamless transition. Because once again, if you can get more specificity from the source of the documentation at the point of documentation, you’ll drive improved processing, improved efficiency through the backend. And the computer assisted coding solution are really meant to provide that first pass that then the coders can work from, edit and submit.[pullquote] if you can get more specificity from the source of the documentation at the point of documentation, you’ll drive improved processing, improved efficiency through the backend. And the computer assisted coding solution are really meant to provide that first pass that then the coders can work from, edit and submit.[/pullquote] So that’s the full workflow of the two things that relate to ICD-10.
Now if you move further down to you next question around Meaningful Use, clinical language understanding (CLU) is also very appropriate here. So the first power addressing Meaningful Use from this perspective, from the clinical language understanding perspective, we have the best-in-class eScription technology platform for transcription. We process about a billion patient records a year through eScription across the US. The workflow in that scenario is that the physician picks up a phone or uses his iPhone and we have a digital dictation recorder and he quickly captures the dictation and sends then, based on the patient, to transcription. eScription platform leverages speech recognition and a modeling engine that pulls that document and presents the transcription with a first pass at the final report, they edit, they QA and they submit that back to the physician for signing in the electronic record signing queue. Upon that submission, we now run that document through clinical language understanding (CLU) and what that produces is a HL7-CDA level 2 document that now is appended to the transcribed final report.
So when it goes back to the EHR, it’s in that format so that they can pull the structured information out of that report that is needed as they’re populating fields for Meaningful Use.
HIT Consultant: From an integration standpoint, does it easily integrate with all existing EMRs for a hospital?
Carina Edwards: Yes, so that’s the best part. So we’ve participated, we’ve been a long-time member of the Health Story Initiative and that defines the HL7-CDA standard, and so as we move to Stage 2 Meaningful Use, EHRs need to be able to consume and reconcile HL7-CDA Level 2 documents. So, it’s a great win-win for the industry. The best part here is that we let physicians use the clinical capture workflow that’s most efficient for them. Be it Dragon Medical on the front end, directly into the electronic medical record or leveraging eScription to transcription on the backend to get their documentation in. But both coming in with both structured, end full physician narrative. So you’re not losing the detail of the patient story but you’re still gaining the structure required to meet all the different regulatory compliance.
HIT Consultant: Now you mentioned the partnership with 3M. Just briefly discuss that partnership.
Carina Edwards: Certainly. 3M is a very strategic partner to Nuance. They are a leader in the encoder business across the US. The relationship is multifaceted. We’ve jointly developed the MD Assist solution. So, what we’ve done there is that 3M’s proprietary knowledge is the amount of queries that they provide for their clinical document improvement program (CDIP). And we’ve taken those query sets and combined it with our clinical language understanding technology (CLU) and our base platform: Dragon, eScription, Dictaphone Enterprise Speech and then 3M’s Chartscripts platform. And now, on any one of those platforms the combination is the MD Assist solution. When you’re using one of those foundation products, MD Assist is sold as an add-on and it combines 3Ms knowledge and queries and we’ve automated that into what is understood and gleamed from the CLU entrance. As the example I gave you earlier, as the document is produced and now it’s texted in a notes fields, clinical language understanding extracts those facts and we run that against the 3M rule set and then we prompt the physician based on that knowledge. So that’s the first part of the partnership was this co-development of MD Assist.
The second part of the partnership is just the embeddedness of leveraging our technology within their inpatient computer assisted coding product so the 360 Encompass, inpatient CAC product is powered by Nuance’s clinical language understanding engine. So we have joint RED teams that work and we have a joint office that drives all different vectors of the partnership.
Part 2 of this interview will be posted soon.
About Carina Edwards:
Carina is responsible for Nuance Healthcare’s marketing strategy and has direct line of authority for managing the solutions marketing efforts of the individual lines of business including HIM, Diagnostics and Dragon Medical. Prior to joining the company in January of 2011, Carina was the Vice President of Marketing and Product Management at Zynx Health. In this role, Carina transformed the marketing and product management capabilities for the Corporation, as well as redesigned the infrastructure, organization, and governance to achieve the organizations aggressive growth goals. Prior to Zynx Health, Carina held global marketing, product management, and business development leadership roles at Phillips Healthcare, Sapient, and Impact Innovations Group. Carina holds a Master of Business Administration (MBA) degree from Boston College, as well as a Bachelor of Science (BS) degree in Management Information Systems and Decision Sciences from George Mason University.