JESSICA
Hey. Welcome to Inform Me, Informatics, the podcast all about public health informatics. I’m Jessica Hill. In this episode, I’ll talk with Mark Conde, who is the Assistant Dean of Information Services at the Rollins School of Public Health at Emory University. Mark also serves as the Assistant Director of the school’s Executive Master of Public Health Degree in Applied Public Health Informatics. In addition to this academic work, Mark serves as a board member of the Joint Public Health Informatics Taskforce, also called JPHIT, and is the past chair of the informatics committee of the Association of Public Health Laboratories or APHL. When someone has worked with so many organizations with so many acronyms, you know they’ve been around public health for a while, which made me curious to know how Mark got started in informatics.
MARK
Well, hi. I’m Mark Conde. It seems like I’ve always been involved with informatics. I started my career as an organic research chemist, and there was a computer actually sitting in the corner there that no one wanted to go near. And someone asked me, “If you do anything with that, we’d like to actually integrate it with this instrument.” Sure. Started playing with it and I’ve really been hooked on the idea of computers enabling what we’re trying to do in science and practice ever since then.
JESSICA
In public health, we talk a lot about data quality and how critical high-quality data are to our work, but we don’t always get into the details about what this term actually means. I asked Mark how he thinks about data quality.
MARK
If I’m in a laboratory and I’m entering a date of someone’s birth along with the specimen and I do it incorrectly, or if I use a format that’s not there, or I don’t enter the date of when the specimen was taken in a different example, I actually have now degraded the data quality at the most common definitive level. So if I don’t really know when that specimen was taken on in a date standpoint, I don’t really know the age, you know. I can’t use it to answer questions later. And then there really is sentience of the idea of data quality. It also has to do with the applicability of the data in that, what I’m collecting, can it be used to answer a question on public health process or practice?
Data quality is also related to process. It’s information that flows that has to do with enabling or driving what I’m doing in my science or in practice of public health. And again, if you fail from the standpoint that the data components aren’t there, they’re not properly, say, given a name so I can recognize what the attribute is, perhaps, or I’m not using standards like a codification, or a vocabulary, or an ontology for consistency in the data. You can quickly see that data quality erodes in its ability to become information and then actually be useful in what we’re trying to do in practice, in our process, or what we’re trying to achieve in the end result.
JESSICA
In your experience, how much of the data quality and the systems that we put in place to ensure it, how much of that is done beforehand, that you’re planning the system, and how much of it is like, “This is the data we have, we’ve got to do the best we can with it?” What does that balance look like?
MARK
You make me smile because this is sort of the classic challenge of our day. Traditionally, data would be kind of collected and then figured out later. In our statistical methodology, some of them are designed to make sure that we get down to a set of usable data in whatever pile we get. So that’s the latter dealing with data problem. But more and more, what we are doing and where informatics is making many inroads is building the data quality at the beginning, when it’s created, when it’s derived, when it’s entered, when it’s translated from an analog manual world to a digital one, when it’s traded between systems. And that’s why we have standards, like one code for assays in labs, like SNOMED codes for, you know, results indications, CPT, ICDs 9 and 10, and it goes on for days of silly acronyms. But that ensures that we’re getting the data quality at the beginning. And more, and more, and more, our advocacy, our design of systems, our paying attention to the way data is collected in the beginning, teaching everybody about this idea of data quality, they realize the relevance. They take on the responsibility of making sure data quality is there in the beginning when it’s created and drive that all the way through to when we need the information for whatever public health practice we’re trying to achieve.
JESSICA
So something that I learned through one of our past episodes where I talked with Dr. Marissa Fond from FrameWorks Institute about their research around communicating about informatics, was the importance of highlighting the role of the people, that this doesn’t happen magically, and it’s not just sort of a bunch of events that come together, that there are people there, informaticians, that design this. So I’m trying to…I’m channeling Dr. Fond right now to ask, like, so what is the informatician’s role in ensuring that high-quality data?
MARK
Well, a lot like we already talked about, the informatician is continuing to look for those areas where data quality in the beginning of generation or derivation is failing, or data is missing, or they realize that two systems can’t talk to each other. How do we actually change the way they work so that they can talk? But another aspect to the informatician that’s really important, and this is why an informatician and informatics is a dual discipline kind of thing. The informatician starts by looking at what are we trying to achieve? What are we trying to do in the practice of public health? Are we trying to save another 10 million babies in a low-resource country? Are we trying to discover the location of where Zika is starting to arise again?
And then derive backwards from the challenge what is the information, what is the process related to data moving, what is the data I need to answer the questions? So the informatician is not only making sure that that data quality aspect continues to be consistent. They’re going backwards and thinking, “Okay, if we’re going to be successful in answering these questions and driving where the epidemiologist needs to go, I have to make sure that the right data is in place, and available, and correlated, and interoperable.” We’ll answer the question, is it connecting on systems that provide the data? And then the technology layers can provide it in a way that epidemiologists can use it. So an informatician, if you would, has a spectrum of responsibilities these days.
JESSICA
So then thinking about preparing to be able to support those roles and execute those functions, how are schools like Rollins preparing future informaticians?
MARK
A difficulty for us for a while was how do you create an informatician, and you have to understand a baseline passion that these two exist in the people who wish to do that kind of role. One is its dual discipline. You have to be excited about technology, and applying it, and using it to achieve something. In this particular case, our public health goals, our intent, or our outcomes. You have to also choose that you’re going to learn and understand enough about science and practice that you understand where we’re trying to go in using the technology.
So you have to develop skills that are existing in both of those disciplines. So that’s base level one, how do we start teaching informaticians? You have to accept the fact, you’re going to grow skills in both areas.
JESSICA
In both information technology and also public health practice, or epidemiology, in particular?
MARK
Well, in our case, we take classes in every one of the areas, relative disciplines from global health to epidemiology, to biostatistics, to health policy. And you really get a baseline spectrum of the major categories of public health, and in each area, you drill down and understanding, you know, you’re starting to…begin to understand how does an epidemiologist think, what are they doing? How do they use information? And then when you go into informatics courses, we use use cases and examples that come from what’s going on in public health to really think through what the technology and data should be doing and then how does the data and technology really work on the IT or IS side?
We’re not creating IT people. We’re creating people that understand how to work with the IT professionals to drive successful technology to do what we need in public health. And that’s where an informatician is kinda cool. It’s a discipline that grew and realize the complexity on the IT side, growing like crazy, just incredible opportunity. And the same thing with the science and practice side, growing like crazy. When you start thinking about genomics, it blows your mind in what we do in public health. But how do you put that all into one human? You can’t. So this great role of the informatician has been formulated to help connect the two worlds, to be successful with applying the technology how and where and the way we need it to succeed in the practice of science.
And so when we’re thinking, what do we have to teach these amazing people? Well, you have to get some base-level tenets in their training in. One, they’ll have to learn to be systems thinkers, in extension, being critical thinkers. And most folks think that they are. But honestly, it’s really hard. Most of us are reductionists in the way we think. We’re very linear, if you would, in a minimalist way. You have to relate and understand to how other things are interrelated. What is really going on? When something occurred, what else is occurring? What else could occur in time? So we teach them to think that way. We teach them to communicate and learn about requirements. How do you use a classic IT way of gathering requirements but do it in a way that you could have a conversation with an epidemiologist? Diagramming methods, information flow models, and workflow models, and whatnot. So you could have the conversation, both understand what’s going on, explain where data information is moving, and with who, and how it changes over time. So that’s another strong tenet.
There’s another side to the informatician that was almost the third leg. They have to understand the business of technology. And so we spend time teaching not only how do you communicate and talk about the applicability of the technology but how to actually fund the expensive need. How do you actually then sit down and talk about responding to something like an RFP to succeed in what you want to achieve in public health and then have the conversation with the technology folks to build what you need? And so they learn project management. They learn about proposals. They learn about OMB 300s, which are a lot of work.
JESSICA
What are OMB 300s?
MARK
That’s a federal form you fill out to actually justify and explain a technology investment.
JESSICA
And do you have discussions around how to do all this on a public health budget?
MARK
Actually, that’s critical. I like to actually go on there. Most of the examples that we yield, public health is a low-resource world. We talk about ways to achieve so that if you’re doing an information flow model on a napkin is just as effective if I had spent the money to, say, buy something like Visio. It’s a really minimalist version. But we do. We come from that standpoint, and it’s almost a little MacGyverism, if you would, in some of the things we’ll teach. We really believe in using and teaching, even here at Emory, I often use open-source-based tools, that you can go and do a state who doesn’t have a lot of resource and still do something complex and build a surveillance system or whatever.
JESSICA
And from your experience, how do students come to want to study informatics, or how do they learn about it and think, like, “Yeah, that’s the spot for me?”
MARK
I think the neatest source that I see people that want to go in informatics is that they were doing one job or the other. They were technology people, or they were practice folks. Very often, like, program coordinators here are in the research realm, and they were thrown into the data and technology side. You know, they would be working as practitioners to do something in the research aspect, but then, suddenly, they had to create a database. And they had to then understand how to make the data useful. And then they had to work with a biostatistician. And then they had to create something, an application to put on a handheld. And it struck them. I would absolutely love to be able to learn a little more of the technology side.
Again, they’re going, “If I know both disciplines, I could do this job better.” Many people come from the practice side and then start learning how the technology works and how to work with the IT people to be successful. And the same thing with the IT people. I mentioned IT is changing. You can’t just create a beautiful box that works wonderfully, is totally safe, and throw over the wall, and walk away. What if you created that in the light that it was trying to achieve this to save a million children’s lives? And you help them understand one more aspect of that. So learning more about the discipline of the practice hit them too. So they kind of come from two directions to, again, become this dual discipline person. It’s usually something they’re experiencing. They realize technology could have saved these people by knowing more about it and know how to use it.
JESSICA
What are some of those final projects that the informatics folks might work on to help understand, one, what they’re learning, but also that application to kind of the world outside these walls and how they’re working?
MARK
You know, they’re literally all over the place. I’ll tell about one that is working on her thesis right now, and the project is actually really pretty neat. She realized that using telemedicine methodologies and handheld modalities, if you would, of collecting information about early diagnosis or a childhood diagnosis of autism and autistic characteristics might actually be more efficient if we could go from a paper methodology that a clinician actually would fill out and use 25-plus characteristics, just do an initial, you know, assessment by usually talking to the parents. To explore a little bit into her project and see, well, if I had that electronically, and then there was some logic in the background that helped determine early that out of 9 characteristics in that 25, I can determine a diagnosis a little faster and then use that data in a secondary sense to start examining a population of change because I already have it digital. How would that happen? And some of the experimentations already proved out that, in fact, if a clinician is using handheld sets of these characteristic surveys and they have, say, machine language type of manipulation or artificial intelligence in the background, they can achieve diagnosis a little bit faster, with less data and less information.
But the challenges come that the students looking at, is, well, that’s lovely, but most of the people that we try and serve in many of these communities, number one, don’t speak English or in low-resource conditions. You know, not all the clinicians are out there in the field. So how do you start exploring different ways of collecting that data in electronic format and then take advantage of, say, machine language or artificial intelligence diagnostic? So they’re starting to explore, “Wait a minute, what if I used, say, Google Translate and I didn’t use paper, didn’t fill out a form, but I ask questions, and I listen to the answers, and whatever language it was, the translation would take it to, say, the nine variables and then interact when they got back?” You’re changing the way that we collect that data. You’re trying to achieve that efficiency. In the experiment, they saw that go from 25 to 9 variables. So that’s really one example of a fairly complex thing where we’re playing some new technology ideas but trying to get more efficient for early diagnosis and then use the population view in the machine language or in the AI diagnosis, which is coming across from that, to determine faster what’s going on with this person you’re learning about.
One of the things I think we’re learning is that the clinical realm and the public health realm are no longer particularly discrete. And when you think about an example like this, the public health view of a population and all of their characteristics, we may be examining changes or rates, you know, of something going on. We’re examining the population. There’s now actually a new feedback loop where it’s being used, say, in training this machine learning system to be the back-end to the clinician who’s doing the patient-level diagnosis. So there are many, many more examples where this feedback relationship is going on, and informaticians are figuring out how they can happen.
And it’s the realization of the clinician in going, “If I had a population view of what was happening, I might actually catch the progression of this disease that I’m seeing with symptoms coming into my ER, because I’m now getting the data coming from Biosense back in a way that I can compare it against these syndromes I’m now seeing immediately with this patient in front of me.” So there are many examples of where this connection is starting to happen not only to serve what we do from our aspect of public health but now actually making clinical realms more efficient. And in the end, you’re actually paying attention to your population on how can you do different things so that less patients come in and go on for tangents on that. But the point is this interrelationship, in many of this work, is just so cool that it’s happening.
JESSICA
So now might actually be a good time to then talk about some of the new tools and technologies that you see influencing public health informatics.
MARK
In the old days, if you would, probably about 20 minutes ago, if you really wanted to use complex IT, you had to learn to code and learn SQL and complex tools, and whatnot. What’s happened finally is that tools, like visualization tools, a favorite is Tableau, and most everybody knows about, an open-source version is Weave, that I’ve seen our public health folks use. The ability to bring the data in in its rawest form and fuse it together or blend it with other data by just playing with it, not having to go through, say, cold-level gluing it together, and then looking at the data and then kinda drilling through it visually, these tools now allow that to happen, and I don’t have to write SQL. I don’t always necessarily have to do the translation, that I can bring it in just as dirty as it is, and visualize it, and play with it.
JESSICA
Actually, I don’t wanna pass over that point too quickly because I think it’s really interesting. So previously, we would have…if different data were coming from different sources, somebody would have to sit down and code each one, if this happens, then this other thing needs to happen to get it all into one clean set, and that’s not necessary anymore.
MARK
Well, it’s necessary. Yeah, no. It’s good. It’s necessary still on many fronts. We’re not discarding what we call dealing with very structured and figured-out data. But now we’re realizing that we can look at that data using devices like Tableau to examine it, to play with it. And I don’t have to get it, and run it through statistical validity and cleaning necessarily to start seeing what the data can tell me first. I can do that along with, say, manipulating it. But that point of the data fusion, data blending, is also a new pathway for us.
We need to do more of that because the questions we’re asking in public health are so much more complex. We do want to climate data and marry it together with, say, syndromic data and then with confirmed case data and laboratory data. And they’re generally not all linked together nicely. So, yes, in the old days, you’d have to write a lot of programs, a lot of code, do the transformations, glue them together, and then look at them. These new tools are really making that infinitely easier to do. And I don’t always need an IT person to do it. And that’s huge. That’s a big transformation.
JESSICA
I closed our conversation by asking Mark the classic Inform Me, Informatics question. How do you define informatics?
MARK
You know, there are so many great definitions out there around what is informatics, but I think the way I view it today is you’re really thinking about the fact that you’re taking data and information. Understanding process, how things change over time related to the data and information, how they apply to what I’m trying to achieve in public health practice or cross a certain tent or goals. And then knowing how to apply the technology to drive and make available those aspects, like data and process, and whatnot. But the true intent is that we’re taking those technology-related things and then making sure, in informatics, that it supports what we’re trying to achieve as an outcome. So informatics is enabling our ability to use technology to achieve better practice.
JESSICA
Thanks again to Mark Conde for talking with us and for sharing his informatics experiences inside and outside of the classroom. I learned a lot about data quality, and it was super cool to hear about some of the current projects happening at Rollins.
Inform Me, Informatics is a project of the Public Health Informatics Institute and the Informatics Academy. You can find out more on phiii.org or follow us on Twitter, @PHInformatics. And, hey, if you like the podcast, please write us a review on iTunes. That helps people find out about our podcast, and more importantly, that helps people find out about public health informatics.
This episode included songs composed by Kevin MacLeod. Many thanks are also due to our production team, especially to our producer, Piper Hale. Piper is to podcast as informatician is to quality data. I’m Jessica Hill, and you’ve been informed.
BUTTON
JESSICA
How do you define informatics?
MARK
It’s a lot of fun, and you should go get an MPH in it.
JESSICA
Piper.
MARK
Yeah.