Sramana Mitra: What are the open problems in your domain?
Drew Ivan: Interestingly, interoperability is a problem that will never get solved. It is not something that is broken that we have to fix and once it is fixed, we need to move on.
Interoperability is almost like a characteristic of the healthcare system because as we start inventing new types of software and new types of data that we want to exchange between software, it drives the use for interoperability.
In a way, that is good for business. We will be working on this problem forever because we are always trailing the leading edge. As software vendors, think of new uses around analytics, AI, and machine learning there is a drive for exchanging more types of data in different ways in different systems. It means more problems for us to attack.
The ones that we are looking at now involve how to get data between organizations properly and aggregating them in a way that is suitable for consumption by business intelligence programs. These are what the healthcare systems are starting to implement.
We are finding that the structured data of the past is a good start for some of these new analytics initiatives, but it doesn’t necessarily go far enough. We need to bring in the unstructured data such as the pre-tech notes that the doctors are typing. We need to figure out something to turn that into something that machine learning can operate on. This is the next hurdle that we have to tackle coming up in the next few years.
Sramana Mitra: How does Postman, Neilsoft, Informatica, SnapLogic, and this class of companies play in your universe?
Drew Ivan: That is an interesting question. There are a lot of different types of interoperability. First, there’s the ETL style of interoperability where you take data out of one database and you may change the format and then you load it into another database.
Second, there is the API style integration where the systems expose their functionality through APIs and then other systems can call those APIs directly.
Third, there is the messaging style of integration. This is the one that I have been talking about, because it’s where our products are most widely adopted. If a system needs to alert other systems of a change to the data, for example if the patient is admitted, the EMR will collect data about that admission and then send an admission message to those that are interested in receiving the messages.
Those messages need to happen in a certain order. They need to guarantee delivery. If a system is down, you need to retry until that system comes back up. There is a chain of custody that is relevant to messaging interoperability that is not seen in those other types of interoperability. Those are the three styles of interoperability, broadly speaking.
We have a perspective on each of those types in healthcare. Healthcare has heavily adopted the messaging style of integration. It’s where we do most of our work. There are use cases for the ETL style of integration. We do that in our product as well.
The API style of integration is just emerging now in healthcare. It has been around for a long time, but it hasn’t seen wide adoption in healthcare until recently when the US government bodies created some rules to say, “You must expose patient data through APIs.”
Now, the API style of integration is suddenly popular among healthcare providers and payers in the US. They are lagging behind other countries, but now with that mandate, we are going to see fast adoption. We have products that target that integration style as well.
Sramana Mitra: Thank you for your time.