I read with great interest a piece today that came to me via my inbox. It was from Health Data Management and the article was titled, ‘Learning Health System’ Requires Interoperable EHR, PHR Data‘.
The article concerns a new report that suggests that for the healthcare industry to deliver on its promise of, ‘learning health system’, an entirely new level of information interoperability needs to take place across “personal health records, personal health devices, patient collaborative networks, social media, environmental and demographic data, and “the burgeoning data streams that will soon become available through progress in genomics and other ‘omics.'” I find this very interesting. At Gartner, we have debates on what interoperability means periodically – we just had one a few weeks ago. And while we might agree on a definition, the “how” it is to be achieved remains elusive. History and experience tells us how to integrate two or more systems that had different designers. Though the data models maybe different, we can map the two (or more) models to enable information to be passed from one system to another. That’s easy. The problem manifests itself when:
- The number of point to point connections becomes cost prohibitive
- The need for more open, share-able, and more easily consumable (i.e. due to unpredictable requests for information) become too great
The healthcare system is looking at the second of these issues. In order to facilitate flexible, agile, even open information exchange, the semantic model needed for the analysis needs to assume a high degree of consistency with the constituent data stores and their semantic models that are being used. If this were possible, health care workers could connect information much more quickly than they do today, and they could also use new tools to discover new insights that elude discover today. This concept of information interoperability or more precisely semantic interoperability , has proven extremely hard within a single organization in most industries. And that is even when the CIO has total control of the systems in question! So think about the chances of making this work across an entire industry!
Some years ago we coined the phrase the “semantic enterprise” to capture this simple idea of massive information interoperability across the firm. This new level, called for in the HealthDataManagement article, is like applying the semantic web to an entire industry: a semantic industry. This would result in a semantic grid; an autonomous trading grid of healthcare information. Sounds great. But why is the “how” so difficult? Whenever this issue comes up in conversation in a research meeting, there tends to be two threads:
- The technologists among us who believe that API’s and RESTful services can cope with or mitigate the challenge of inconsistent semantics, and
- The governance focused among us who believe that the problem is not about the technology that connects the silos at all, but is about the underlying methods used to maintain the maps or transformations between the semantic models.
This actually neatly fits one of our Gartner Symposium 2014 Keynote ideas – The Digital Humanist Manifesto. Restated:
- Digital machinists believe that better technology will automate away the complexity, and
- Digital humanists believe that people who set the roles and polices need to work differently together
Neither is right on its own. Both are needed. But in my humble view the humanists have several secrets up their sleeves that the machinists have yet to contemplate. One I will mention here is this: for an organization to achieve effective information interoperability that said organization needs to understand, model, and govern, a concept I call process interoperability. Without process interoperability designed into the architecture, no amount of information interoperability will persist. And what is process interoperability? For this I have to lean on Peer to Peer (P2P) computing. Several years ago my research into P2P computing gave me a method to re-think a business process such that what was once previously thought of as two distinct enterprise processes connected or integrated (what we might call B2B). This re-think created a wholly new process that is shared by the two organizations. Such a change in the process design leads to a fundamental change in the application design. At the heart of this is what we call a Multienterprise data model. This then changes the nature of how the applications are connected (or integrated). That, my friends, is what the healthcare industry needs.
Now we can all sit back and watch how this plays out…
Sent from my iPad using Mail+ for Outlook. Pls excuse any typos.