CloverDX customer Diameter Health’s technology refines raw patient data into high quality, actionable information at scale for healthcare organizations.
We spoke with Diameter Health CTO Harvard Pan to find out more about how Diameter refines data for customers, and how CloverDX has helped make their data processes, and their customer onboarding, more efficient.
CloverDX: Hi Harvard, thank you for talking to us. Can you give us a little bit of background as to who Diameter Health is, and what the company does?
Harvard Pan: The short answer is that Diameter Health aims to be the standard for health data optimization - to make clinical data usable and actionable by healthcare organizations. So, much in the in the same way that Google makes data available to the general public - we're not a search company - but we aim to provide that same level of usability with our data within the health care industry.
And how do you help your customers manage and work with their data?
We work within multiple markets across healthcare that use our software to make better decisions. So, for example, take life insurance. They want to underwrite an applicant for a life insurance plan. They have access to a lot of data, but they have no tools to manage it other than someone poring through hundreds of pages of unorganized, duplicated content.
Our technology distils this raw data into a single longitudinal view of the patient that can be easily reviewed, highlighting specifics that underwriters care about – medication reconciliation, problem history, etc. And we can organize that data in a common format to support analytics and multiple use cases.
Could you tell us about the project that you initially brought CloverDX in for?
The primary reason that we licensed CloverDX was that we were developing a product to orchestrate the movement of data from our system into a delivery location for the customer.
We would get a patient list, and based on this patient list we wanted to make calls against our technology’s APIs. We would then get the output files, generate them, and put them in the customers location. And this would have to be scheduled on a regular basis - so it could be weekly, could be daily, whatever the requirement is. But what we found was that, as we were developing this product over a number of months, we still had some challenges. For example, consider scheduling. Building your own scheduling is a very complex effort, and we estimated that it would take quite a bit of effort to do it properly.
So what happened was I took CloverDX and created a prototype of what the team was tasked to do.
And it took me about 2 weeks, and with that prototype we were able to demonstrate the time-to-value utilizing CloverDX technology would bring to Diameter Health. I mean, it wasn’t productized - I want to be clear it was a prototype - but it helped prove that CloverDX was capable of doing the orchestration we required with very minimal uplift by engineering.
What were some of the other challenges that you thought CloverDX could help address?
One of the challenges the industry faces is a standard in medical data called HL7 version 2. Structurally, HL7v2 documents are complex pipe-delimited flat files. Even though it's a published standard, everyone implements HL7v2 as a point-to-point solution with significant variability.
The challenge that we faced was to turn this real-world non-standard HL7v2 into something standard, at scale, across a number of customers. You can imagine, you have 10 providers that are giving you these files, and even if it's the same type of file, same version, there could be variations within it, so we needed a process that could handle the variations in a configurable way.
Could you tell us a bit more about the solution you implemented for that?
Our main question was how do you enable someone who doesn't necessarily have coding experience to configure what an HL7v2 file looks like, transform it into our data model, and then send it to our APIs? And so the project that we worked on with CloverDX dealt with firstly, the transformation of these HL7v2 documents into our JSON model, and secondly, building a configurable interface.
We created essentially a custom language, almost like Excel formulas. An analyst takes an Excel file and defines field mappings, and then the output of the process driven by these formulas is a JSON stream sent to our servers for processing, and then once that was done, reporting and status and other deliverables were connected to it as well. So overall the project was to create an end-to-end workflow to generically ingest HL7v2 documents.
So where does that come in when you’re onboarding new customers?
We have an implementation team that deploys the solutions, but as a part of the implementation process, when a customer brings on a new data source we have a default mapping file that requires additional configurations that need to be made. We have a clinical informatics team that works with customers to make sure that they understand the nuances of their files and then configures those details.
And what difference has this made to your onboarding process?
With the CloverDX solution, we can present our customers with an out-of-the-box solution where they don’t need to spend extra effort and IT cost.
Which is a big win for them, but also a big win for us because we can offer a scalable service which is a new revenue stream for us.
And of course, the whole process of delivering value to a customer is much quicker; we’ve been able to do it within a month. And we have more control over the process.
And this is running in the cloud, right?
Yes, it's running in the cloud. We worked with the CloverDX professional services team to build this out within the ECS (Elastic Container Service). We built Terraform modules ourselves that can be deployed in a containerized format in the cloud.
We’ve talked about how it helps your customers get onboarded more quickly, but as a CTO, what would you say CloverDX brings from your perspective?
My job is a CTO is about identifying the right technologies to use, but it's obviously also about how it ties into our customers’ deployments. So CloverDX to me - its value is in reducing the friction needed to be able to roll scalable solutions out to customers. It is great at orchestration. It's strong in its ability to extend out-of-the-box functionality.
I can't tell you the number of times that customers have come to us with what I would consider to be strange requests, and I would say that for most part the thought process is always ‘How much of a development effort is this going to be?’ I have a lot of confidence that with the extensibility of CloverDX that that development effort is going to be minimal.
One example is a customer who needed to limit the zip files we send them to 5,000 files each. We had a number of output files coming out from the system and they didn't want individual files. It would blow up their logging mechanism. They wanted us to put exactly 5,000 files into every zip file and then upload that zip file over. Through extending a CloverDX component, we were able to do this in maybe 10 lines of Java code.
In another example, we had a different requirement where we needed to be able to add certain fields into MongoDB - and MongoDB's query language doesn't support exactly what we needed. Again, this was not that big of a custom component, maybe 50 to 70 lines of code. But we were able to do everything that we needed to build a component in Java using the Jackson library.
The extensibility gives me flexibility to be able to say to customers ‘I can do this for you.’
To see how CloverDX could help you solve your complex data challenges, request a demo