I find the publication of an enumeration of common UCUM compositions to be very useful, and we are considering binding to this list in the Federal Health Information Model. However, to so, we would need an identifier (OID or URI), and we’d like to see it published in a terminology service (e.g., VSAC). Would Regenstrief consider assigning such an identifier? Has anyone else done so?
Is it helpful to know (or did you already see) that PHINVADS has a value set of <span style=“color: #000000; font-family: Tahoma, Verdana, Arial, Helvetica, sans-serif; font-size: 11.2011203765869px; line-height: normal; background-color: #eff5ff;”>PHVS_UnitsOfMeasure_CD named Units of Measure with the OID of </span><span style=“color: #000000; font-family: Tahoma, Verdana, Arial, Helvetica, sans-serif; font-size: 11.2011203765869px; line-height: normal; background-color: #eff5ff;”>2.16.840.1.11418.104.22.1688 ? The description reads </span><span style=“color: #000000; font-family: Tahoma, Verdana, Arial, Helvetica, sans-serif; font-size: 11.2011203765869px; line-height: normal; background-color: #eff5ff;”>Units of Measure based on UCUM std which also includes dimension . </span>
I’m not certain if this is one degree away from what you’re needing, or if it’s a fit. Thanks for the consideration.
Forgive my earlier post; I didn’t realize pasting in content from the web could carry the formatting. Here’s what I was trying to show:
Is it helpful to know (or did you already see) that PHINVADS has a value set of PHVS_UnitsOfMeasure_CD named Units of Measure with the OID 2.16.840.1.114222.11.838? The description reads Units of Measure based on UCUM std which also includes dimension.
<span style=“color: #464646; font-family: Lato, arial, sans-serif; font-size: 15.1875px; line-height: 23.73046875px;”> </span> pb
I have a (perhaps) related question.
I’m sure published UCUM values would be useful, but more useful to me would be a published list of possible ordinal values.
for example, for programming values received, a numeric–like ‘serum calcium’ can go directly into my DB as a number.
then evaluated by my AI
but for ordinal values, one shop may return ‘Positive’ another ‘T’ and another ‘+’
this is quite unpredictable, and difficult for my AI to process.
has anyone considered creating a list of “possible” and “preferred” ordinal values by test?
Hello Dr. Hollington,
It certainly is a related question. Where UCUM would describe units for the numeric values crossing to your database, SNOMED CT is expected to describe non-numeric values. The Qualifier domain contains descriptions for non-numeric answers, such as negative, positive, equivocal. Our clients provide a listing of the canned answer values they attach to laboratory assays, and the mapping to SCT is then attached in their LIS, for sending in the outbound HL7 OBX record. The same translations are made for genus species of organisms found in cultures, so that local abbreviations for MRSA, O157, etc are mitigated.
I hope this helps.
It does help, Pam.
thanks for your time. the issue I have is that we are working on a generic AI that is a client for multiple systems. so having each partner system submit a list of acceptable values is at best cumbersome. Tell me about this ‘qualifier domain’ is that a SNOMEDCT thing or an HL7 thing or a LOINC thing.
I still believe it would be useful for some standards group to assign or manage preferred ordinal values for each test. I am certainly willing to submit what I have–but who might I submit them to?
Hello Dr. Hollington,
SNOMED CT contains the qualifier domain that I alluded to in an earlier post. As to your second question, I am not certain which standards group would be appropriate. My first thought is the AACC’s Harmonization Initiative. Here’s their URL if you want to investigate further. https://www.aacc.org/global-health-outreach/harmonization
A second thought is that Regenstrief does track example answers for assays, but it is usually derived from the submitter initially. I’m unsure if your list would be revisiting previously developed terms. You can always ask by emailing email@example.com