These articles popped up on my radar this morning: ‘Should We Create Digital Twins To Better Treat Patients?’ and ‘Why the FHIR standard may not get us to interoperability’.
The lack of a common health data language identified in the second article has been ‘the elephant in the room’ for a very long time. Unfortunately, very few people acknowledge the need for a clinical lingua franca as a critical foundation for eHealth. The mainstream view seems to be that messages are/will be enough and that creating a standard language for health information is either too hard or too complicated. Is it really that hard? Or is that just the view of those with vested interest in perpetuating the message paradigm?
The elephant = a universal lingua franca that is common to all digital health implementations.
In response to my question about what Australia was doing to standardise health data at a recent meeting on big data in research, Australia’s Chief Scientist, Dr Alan Finkel explained to a conference of health data researchers that defining atomic health data standards had been the holy grail in health IT for the past 40 years, but concluded that no one had ‘cracked it’ yet. Umm, no, not quite right! Over 20 years ago a little Australian invention, a specification for an electronic health record (EHR) known as openEHR, was first published. It described an architecture for a health record implementation as well as the technical approach to a computable health lingua franca based on a standardised representation of information models, known as archetypes. Barely recognised within its’ country of origin, openEHR has been making inroads into semantic and international cross-border interoperability in Europe ever since.
openEHR archetypes express patterns of clinical knowledge in a computable format – clinicians can understand them and technicians can implement them in clinical systems. They have been in existence since early 2000’s, admittedly in very small numbers in the early days, but numbers are growing and strong semantic patterns have been established, underpinned by tooling and an online community that collaborate on the refinement of the models, verifying them as clinically appropriate and ready for implementation. There is no comparable effort freely available in the world.
The clinical modelling work done in openEHR is similar in principle to that of the HL7 CIMI effort but is far advanced in terms of the range of content, methodology and approach, international community collaboration, and governance and publication processes – the purpose-built online CKM tool is the hub of the openEHR clinical modelling community. The 2018 end of year message from the openEHR clinical program leaders summarises the international openEHR community progress to date - it makes interesting reading. This is not a trivial project, but potentially world-changing community of practice and one that I am extremely privileged to co-lead.
Consider also that the openEHR specs have been around for a couple of decades. In that time we have seen the tech ‘fads’ of HL7 v2 (including Standards Australia’s TR 2964-2010 Representing archetyped data in HL7 Version 2) and CDA come and go, and are currently fully experiencing the hype and excitement of FHIR which will persist until the next technical/implementation paradigm is invented… it is inevitable. Yet, based on the stable openEHR specifications, the archetypes persist over time – vendor neutral, public domain, freely available, agnostic to any specific technology or implementation paradigm – waiting to embrace whatever new technical trends and innovations come in the future. The archetypes will persist and be ready !
As pointed out in the FHIR-related article, above, it is illogical to expect that we can achieve semantic interoperability via limited messages with non-standard content. And if we desire the AI capability for creating and managing innovations such as ‘digital twins’, we need a systematic way of representing all clinical data, not just the content that underpins limited exchange formats for sharing a health summary or test results. This kind of digital advance requires a strategy that embraces the full breadth and depth of all clinical knowledge. Until we have a comprehensive standardised and computable clinical language semantic interoperability, AI, precision medicine, serious and safe big data etc will only be operating in a severely limited capacity.
Australia’s health IT leaders, along with many other national programs, are ‘on fire’ for FHIR. The majority live by the latest iteration of the old IBM mantra ‘You won’t get fired for implementing FHIR’ – all puns intended and not original! Yes, FHIR does provide welcome, quick wins. I’m not denigrating the value of FHIR – it does what it is designed to do well and is allowing for further incremental progress towards interoperability.
My message to you is that FHIR alone is not enough. And the fact that it is being embraced by any or all of the technical giants doesn’t change this reality. To leverage the benefits of FHIR we also need technology- and vendor-neutral standardised atomic, clinical health information. Maybe even more importantly we need an approach to data standards that successful engages with the clinical community. This is what HL7 has not yet been able to crack and this is where I think that strategic collaboration would benefit both organisations.
It is no secret that each openEHR implementation relies on the use of other standards, each selected as the best one for the job! We want to, need to, work with HL7 and other standards. None of us can do this alone. It is almost universal that HL7 v2, FHIR (and even CDA in the past) are used alongside our archetypes. Yet the HL7 community, almost universally, seems blinded to the value that the openEHR semantic standardisation effort offers. Rather than engage and collaborate, over and over again the HL7 community have preferred to reinvent the semantic wheel – we’ve seen it with CIMI, now again with FHIR.
Change is well overdue. We need to abandon outdated ‘not invented here’ thinking and collaborate across standards bodies. A common health data language standard is an essential foundation for digital health disruption. Our eHealth mantra should simply be 'the right standard for the job'. Stop the pointless, expensive health IT standard religious wars! Let’s try some common sense and adopt a collaborative model so that the entire eHealth ecosystem benefits.