Are you Content With Your Content?

Charlie Harp, CEO (00:04):
Good afternoon everybody. Good morning everybody. Thank you, Joe. Many people don’t know that many, many decades ago in a lab near Chicago, there were folks that decided to embark on a bold experiment. Many said they were crazy to do this, to create the perfect clinical informaticists in a lab genetically that had never been done before. But ladies and gentlemen, I’m pleased to tell you that we have with us today my good friend and esteemed colleague, Victor Lee, the VP of Clinical Informatics at Clinical Architecture, who’s going to talk about or actually ask you the question, are you content with your content? Content with your content? Victor, over to you.

Victor Lee, MD (00:52):
Okay, thank you Charlie for that embarrassing introduction.

Charlie Harp, CEO (00:56):
You’re welcome.

Victor Lee, MD (00:58):
Hello to everyone in the audience and on the recording. Happy Pie Day. If you’re watching this recording, it might not be pie day anymore, but if you celebrate that, I hope you get some good pie. I’m pleased to talk to you today about clinical content. As you may know, clinical architecture is a software company and we’re really good at what we do with software and sometimes that overshadows some other really high value offerings within the company. Namely, we have in addition to World-class software, we have professional services and content that sometimes gets overshadowed just by our magnificent software. But I’m going to focus on content today, and I’m really happy that I was given this opportunity because one of the things I often say is that you never want valuable assets within the company to be the world’s best kept secret. So I’m going to shed a little bit of light on content today, and I’ve created kind of a framework for how I’m going to discuss the various content offerings at clinical architecture.

(02:13):
One aspect of how we can categorize clinical content is whether there is additional payment required to access the content because there are some proprietary content artifacts that do require payment. The ones that are publicly available, we have an acquisition and distribution team that puts it in our some medical subscription portal. That’s the data quality software platform. And that will go into the content portal section of our subscription portal. And then there are other content artifacts that do require additional payments. And I’ll go through some examples of these, but I just want to talk about the categorization. So the ones that do require additional payments are placed in the content marketplace. And then another way to categorize different parts of content and terminology are who the steward is. And so there’s third party content. So for example, you might’ve heard of SNOMED, LOINC, RxNorm, those are the steward are other organizations that author and maintain this content.

(03:20):
And then there’s stuff that is authored and maintained by clinical architecture, and there’s a good chance that my team is the steward of that content and we publish it on a regular basis and also put that through the subscription portal. So the focus of this, and there are many, many of these content artifacts. In fact, I named some as example SNOMED, LOINC, RxNorm. There are more than 450 packages of content. And this is the stuff that I don’t want to be the world’s best kept secret. And so this is kind of the focus of what we’ll talk about today. I have this two by two grid with the framework that I just presented in the previous slide. If we look at the first box in this two by two grid, we have content where the steward is a third party, and you can find content in the content portal section of the Symedical subscription portal.

(04:22):
Remember, there’s more than 450 of these packages. So this is just a select number of examples. But for example, LOINC, is the logical observations, identifiers, numbers, and code. The steward is the re Regenstrief Institute. And if you’re familiar with the United States core data for interoperability, these are ONC recommendations for how we should communicate patient data in an interoperable manner. And so U-S-E-D-I would name an applicable vocabulary standard as the recommended way to exchange certain kinds of information. So LOINC would be used for example, for the interoperable exchange of observations, which include laboratories and other kinds of patient observations. We could probably do an entire session on LOINC, but that’s not the focus today. But LOINC is one example of third party content, and we have a wonderful team that’s actually overseen by Diana Kaufman who makes sure that we fetch the original LOINC content and package it and make it available so that we can do all the data quality things that we need to do within some medical.

(05:35):
So LOINC is one example. RxNorm is owned and operated by the National Library of Medicine. They come out with weekly updates of drug information. And so that’s another content asset that is named within the U-S-C-D-I for the interoperable exchange of drug information. So you have a lot of patient information laboratories results, medications, which should be expressed through RxNorm for interoperable exchange and other use cases. And then SNOMED CT, it’s the systematized nomenclature of medicine, clinical terms that is owned and operated by SNOMED International. And so again, our acquisition team will go fetch the raw information and package it in Symedical. These are all ontology. So they’re not just terms, they have descriptions, relationships, there’s a very rich set of information in each of these packages. They all have their different use cases. These I could go on. There’s ICD-10-CM, the United States Clinical Modification.

(06:50):
There’s a whole bunch of assets that can compliment your data quality initiatives. So these are just examples in this box. There’s probably hundreds more that we don’t have time to go through, but these are examples of third party content available at no additional cost. And in this box, third party content, which is available through the content marketplace portion of our subscription portal. These are ones where there are certain licensing requirements in order for people to use them within some medical. For example, CPT also known as current procedural terminology is owned and operated by the American Medical Association. And so what we will do to enable our clients to use CPT within some medical is that we simply ask for proof that they have licensed CPT directly from the AMA we verify with the American Medical Association. And if they say yes, this client of yours has indeed paid for CPT and can use it within some medical will activate it for them. So it makes it easy for people to use these terminologies for their data quality initiatives. Another example might be drug information and CPT their procedure codes. Medi-Span, as an example, includes drug codes. There’s a variety of these types of third party content assets.

(08:29):
We don’t monetize any of this. We simply make this available through some medical for our clients to make effective use of them within some medical. Okay, so in this next box, this is sort of the right column is what I’m really passionate about because my team is the content steward for the artifacts in the right hand column. So clinical architecture is the content steward, and I’ll spend a little bit more time talking about these things. On the right hand column, there are content artifacts that we make available to our Symedical subscribers at no additional cost beyond just their somatic license fees. So we provide content to solve certain types of problems. And one example is the Allergies and intolerances catalog. This is a catalog that many of our clients use as a normalization target for the interoperable exchange or possibly analysis of drug allergies and intolerances.

(09:41):
So it includes things like substances that people might be allergic to, medications, drug classes, I mentioned substances within substances. There are many different kinds of substances such as food substances, non-food substances. People can be allergic to things that are not considered food, like maybe food additives like saccharine. So patients may be allergic or intolerance to any number of substances. And so these are all represented within the Allergies and intolerances catalog. So for interoperable exchange of allergies and intolerances, many of our clients will perform mappings to normalize their local allergies and intolerances to kind of a central standard. Okay? And there’s also external codes so that if you needed a cross-reference over to SNOMED, Unicodes, any number of external codes, those are all present. There’s a lot of attributes that are supported within the allergies and intolerances catalog. Again, all available to our subscribers at no additional costs.

(10:50):
We have a clinical architecture also supports a units of measure, catalog and model. This is for the intra conversion of units of measure. There are also API functions that support this. So if we wanted to convert values from one unit to another, for example, if I wanted to convert a gram to a thousand milligrams, we provide the ability to publish this unit of measure model to a runtime so that these conversions can be done through API access. Okay? So unit of measure catalog. If that’s a problem that any of you are trying to solve, we have a unit of measure conversion application for that or content.

(11:45):
And then in this final box, we have a variety of content artifacts that my team develops and maintains these are available for an additional fee and that’s why they’re placed in the content marketplace. Just within the time that I have here, I’m not going to go, we could do an entire presentation on each one of these artifacts, but a align stands for the aggregated lab information grouping nomenclature. That’s a Charlie acronym that we’ve created. And basically the problem that we’ve understood in the industry is that people have been looking for groupers. They want to find common things, common laboratory test results that can be grouped together. And some of the use cases for grouping labs might include graphical trending. So for example, an organization might want to take a series of patient lab results and trend them together, but you can only trend lab results if they have the same, for example, component.

(13:00):
If the method of analysis is the same, if the expected units of measure are the same magnitude, it doesn’t make sense to group things together when they are reported with different units of measure, for example. And so we might need groupers to say these things are similar and therefore they can be graphically trended together. There are also additional use cases for analytics where we might want to have groupers and Align provides those. So kind of like a value set, if you will. If we wanted to run some analytics on patients with hemoglobin A1C or other labs, albumin and I care about the difference between a serum versus a urine albumin, those are not the same thing and they should not be grouped together. So Align provides those types of groupings. There are also certain electronic health record systems that have functionality where based on a given lab grouper, they can invoke a certain set of lab results within a note authoring workflow.

(14:09):
So as a physician might be documenting about a patient encounter and they don’t remember what the last hemoglobin A1C was, rather than leaving their note authoring workflow and then looking up labs, certain EHRs have the ability to leverage the groupers and invoke the latest lab values for a given component and have that brought up directly in the workflow to make it easier for documentation to occur. So those are some example use cases for Align laboratory groupers. And Align is not dependent on LOINC, Align, it can be complimentary to LOINC. But one of the things, if you were thinking about lab groupers, why don’t I just turn to LOINC and get lab groupers? Well, the answer is that you can’t because LOINC does not have groupers. And we’ve actually had some conversations with the re Regenstrief Institute about whether they plan to support lab groupers, and at least at this point, they have no plans to do so.

(15:17):
And therefore this is a solution that is complimentary to LOINC organizations that have lab values, lab laboratory concepts in their laboratory compendia will want to map to align, and often they’ll have link terms associated with their local lab terms and we can use those as reference maps to help connect their local terms to align. So we’ve thought about kind of how this could be implemented and certainly talk to me if you want to hear more about Align. Okay, this next one in this box is CA Elements. And if you’ve attended a prior conversation, a prior data quality theater presentation with Joe Bommel and myself, we talked about value sets and we made some reference to CA Elements as well as Jim Shalaby, I apologize. But we have a value set offering. We call them elements within Symedical, but basically we have a collection of more than 2,400 value sets that span many clinical domains, conditions, medications, laboratory tests, procedures, value sets can be used in many versatile ways, whether you’re trying to do analytics, clinical decision support if you’re trying to, we had a recent discussion around electronic laboratory reporting.

(16:54):
We need value sets to understand tests for reportable conditions. So any sort of analytics research effort, anytime you’re trying to get intelligence from your patient data, you may need value sets CA Elements, also known as Clinical Architecture Elements set foundation is an offering of more than 2,400 value sets out of the box. We have clients that are using them for a number of use cases. We’re also because value sets are custom made for different use cases, we are very interested in hearing from our clients about their use cases for value sets. And we really enjoy working with our clients to provide additional value sets that might be scoped a little bit differently than how we’ve created them. But we generally take a stand and make very comprehensive value sets and are very interested in client feedback in terms of crafting our offering to make sure that we’re meeting everyone’s needs within CA Elements.

(18:01):
There’s also the ability not just to take the value sets that come out of the box from clinical architecture, but we have some very savvy tooling that our clients can use to further extend or to entirely customize their value sets. And we support intentional as well as extensional. And when I say intentional, I mean intentional with a S not with a T, meaning that we can have rule-based value set definitions for scalability of value sets. And so in addition to the CA Elements content offering, you may be very familiar with the software tooling that allows for the additional customization of value sets. So we have a comprehensive solution to meet your value set needs as well. And the last item in this box we call inferences for short. Our official brand name would be the Advanced Clinical Awareness Suite. Basically, these are rules-based logical reasoning capabilities. So we have the ability to extend our expertise with terminology management and our understanding of models and ontologies relationships to bring inferencing capabilities.

(19:24):
So we can interrogate structured patient data and reason over the data based on configurable rules and then return assertions based on what these rules have discovered. So as an example, we’ve done implementations where we have helped organizations detect undocumented conditions. So if you’re doing a population health initiative, or perhaps if you’re trying to perform risk adjustment and maximize HCC capture hierarchical condition category capture, you might be very interested in completing out the information in a patient record. So if a patient has undocumented diabetes, for example, it might be that the inference rule logic would identify the fact that there’s an elevated hemoglobin A1C, or perhaps the patient is already being treated for diabetes and you see insulin and sulfonylureas and other drugs that are evidence that the patient actually has diabetes, but diabetes mellitus is actually missing from the problem list. We can use inferencing, which are actually dependent on value sets to understand if I don’t see diabetes, and if I do see these medications or these lab values, or perhaps there are other conditions that are complications of diabetes like diabetic ketoacidosis.

(20:52):
These are compelling indicators that the patient might actually have diabetes. In an earlier presentation today, we had another example of inferencing, which is the electronic laboratory reporting use case. So if we find that there’s an HL7 message that contains a particular laboratory test, and let’s say it was positive for a SARS-CoV-2 that could inform the data pulse solution, we just talked about that in another presentation that might satisfy reportability criteria for a given jurisdiction. And therefore this logic can help an organization acquire insights about patient conditions and therefore we can act on them accordingly. Okay, so these are just examples. Again, there are more than 450 packages of content. Each package may be rich with lots of conceptual terms, relationships, descriptions, and this is the last slide. I just wanted to provide an opportunity to get a little bit of insight in the content that thrives some of these data quality initiatives. So thank you very much.

Charlie Harp, CEO (22:12):
Can I chime in real quick, Victor? Absolutely. So one of the things I want to point out is part of the reason why clinical architecture does this is because I think getting content, getting the right content, current content is a really big part of quality that people sometimes don’t think about. And part of the reason we did this is we noticed that a lot of the people out in the industry, and I’ve been doing this for 35 years, they getting the content you have to go to maybe 50, maybe 60, 70 places to grab the content. You have to create custom loaders to pull the content down and to get into a database and make it work. Sometimes they change the format of the content. So a lot of the stuff you see that’s in the third party content bucket is because what we wanted to do was eliminate that issue.

(23:02):
That should not be an issue that stops you from having accurate results, good data, current versions of standards. We should be able to streamline that for you and we do. The other reason we do it is when things change, we just take care of it for our clients. And the truth is, when you look at the first column, the first column is because a client comes to us and says, Hey, we need this content. Can you go get it for us? Once we get asked that question, we go get it. We add it to the content portal, and from that point forward, it’s our responsibility to make sure that it’s always updated, it’s always correct. If we see any issues, we go back to the content source and we tell them and we fix it before we give it to our clients. The stuff that’s on this side of the column is really when a client has a need around analytics or something and we look around and we just can’t find it anywhere.

(23:55):
The thing they’re trying to do, the thing they’re trying to deal with, if you look at the allergies and intolerances, allergies, when it comes to normalization is for a long time it was the wild west because there are so many different terminologies that could represent an allergy. So rather than tell clients, oh, well you combine SNOMED and RxNorm and this and that, what we decided to do is we’ll just go ahead and we’ll do it for them. We’ll do it for them, we’ll put it in one place, we’ll make it available. Same thing with units of measure. People think that UCUM is a terminology, but it’s not. It’s a nomenclature. So what we try to do is create a terminology that represents the nomenclature so people can use UCUM, essentially like a terminology and also have some of the programmatic convenience of conversions and things like that.

(24:41):
With the stuff down here, it’s the same thing. We have clients that say, Hey, we’d really like to be able to do these certain things. We go out, we scour everywhere for a terminology that’ll serve that purpose, and if we can’t find it or if we think we can put something together that’s better, then I give it to the best content team in the business and that’s Victor and his group. So I just wanted to give the background about why clinical architecture, a software company did this. And a lot of it is because we’re a software company that caters to terminology. So we just try to take some of the busy work out and ratch the quality up.

Victor Lee, MD (25:19):
It’s the easy button for content, it’s the

Charlie Harp, CEO (25:21):
Easy button for content. That’s true. Victor.

Victor Lee, MD (25:25):
Alright, any questions from the audience?

Audience Speaker 1 (25:30):
First off, fantastic. This was the best display review of what clinical architecture is doing. It’s really helpful together for me. I’m interested in, you mentioned groupers a number of times and in that bottom right hand box, and I’m curious how far you go in terms of whether you actually do things like episode treatment grouping or groupings for consumer cost predictions, that sort of thing.

Victor Lee, MD (26:01):
Yeah, great question. The question was about groupers and in the align, the aggregated lab information grouping nomenclature, we provide two different levels of grouping today, and that’s not to say that this will always be the case, but we listen to our clients and the first level of grouping is one that allows, it’s a very narrow set of groupings. It’s probably optimized for the graphical trending use case that I mentioned. There’s higher level of grouping that includes a broader collection of laboratory results, and those typically support a variety of analytics use cases. So for example, just going into a little bit more depth, a lot of the grouping methodology has been designed around Epic functionality. So within Epic I mentioned the first level of grouping, they have the ability to do graphical trending at the broader grouping level, they are able to leverage the groupings in their analytics tools like SlicerDicer, and the note authoring use case that I mentioned. They can invoke a grouper within a clinical note. So they’ve been optimized around that. Now, the beauty of Align is that if we wanted to support other levels of grouping, we simply just add them as additional attributes to the Align catalog. And so when clients come to us and say, I’d like a different level of grouping, I have a different methodology in mind, we would talk to them and try to understand the problem that they’re trying to solve and see if that makes sense to create alternate groupings. So great question. Thank you.

Charlie Harp, CEO (27:53):
Right. The only thing I would add is when you think about the Symedical platform and Pivot downstream, the other nice thing about having all this content available is Clinical Architecture has a rich set of APIs. So you can take the content, you can enrich it, you can publish it to a runtime, you can publish it into your own repository and you can use it for mapping, you can use it for building your own ontological structures or on top of an existing ontological structure. So it’s really one of those places where the terminologies come together, whether it’s something we build, whether it’s something that comes from a third party, whether it’s something you build, you’re able to kind of bring everything together, create the framework for your universe and deploy that into the world. Thank you very much, Victor. Any other questions? Alright, thanks a lot.

Victor Lee, MD (28:47):
Thanks Victor. Thank you.