I recently submitted a speaking proposal on a practical approach to achieving semantic interoperability, and was shocked to see the response from one of the reviewers. He intimated that the Unified Medical Language System (UMLS) is all that is needed to solve this complex problem. Furthermore, he indicated a disdain for any proprietary solutions. (Somewhat disingenuous, given that much of UMLS is proprietary—SNOMED and CPT amongst others). I thought I would take some time to discuss the common misconception of the UMLS as a panacea, for this is a banner that is commonly raised by well-meaning, but naïve individuals who dream of simple solutions to complex problems. I certainly understand the appeal of the National Library of Medicine’s open source efforts.
Unfortunately, while the UMLS may be an important tool, it cannot be a solution, and this is borne out by its failure to attain this goal despite its creation nearly 25 years ago. Now to be fair, it was never intended for this purpose.
The UMLS is not software. It is a system that was designed for information retrieval. The idea was to get information from different databases and present it to users—a noble goal to be sure.
Real clinicians need to do much more. They need to integrate information and knowledge (not merely data points) into their workflow in order to enhance the care of their patients in an efficient manner.
As I have stated in previous blogs, standards are important. So many of them exist—HL-7, CCR, CCD, ICD-9 and 10, SNOMED, LOINC, CPT, etc. Standards are necessary, but not sufficient, to obtain true interoperability.
I do not profess to be an expert on this subject. I remain a Family Physician dedicated to establishing a practical solution. I would like to share a criticism of UMLS, however, by someone who is very much recognized as an expert.
Dr. Barry Smith is the Director, National Center for Ontological Research, and Julian Park Distinguished Professor of Philosophy, at the University at Buffalo. I highly recommend taking a look at his blog. When writing about the Top 10 Errors To Avoid When Building Ontologies, he specifically calls out the UMLS for the sin of mixing ontology with epistemology. The example he cites is "cardiac output is a Laboratory or Test Result or Diagnostic Procedure”
Having said this, let me be sure to clarify—my goal is not to put down the UMLS— it does not deserve this, nor am I qualified to try. Instead, I wish to promote more enlightened conversations in an attempt to solve problems. Like other complex issues affecting health and wellness, there are rarely panaceas.
I have ranted enough… and feel much better!
