Page tree

2246 View 7 Comment In discussion Comments enabled In the category: Undefined

Development of clinical observables ontology for US domain

Contributors (2)

7 Comments

  1. I am building out a LOINC-on-OWL ontology for consumption bythe community of LOINC users.  Laboratory and pathology are in pretty good shape and I am progressing through a frequency analysis of clinical observables concepts deployed and in use in US healthcare records.  This is the top 20 list from UNMC and Intermountain.  

    LOINC_NUMLONG_COMMON_NAMETotal
    8867-4Heart rate349847977
    9279-1Respiratory rate241742000
    8480-6Systolic blood pressure228900486
    8462-4Diastolic blood pressure226008756
    8478-0Mean blood pressure168467673
    8310-5Body temperature71671221
    3141-9Body weight Measured33262360
    8302-2Body height26433880
    39156-5Body mass index (BMI) [Ratio]24710685
    3151-8Inhaled oxygen flow rate12131789
    9267-6Glasgow coma score eye opening7694651
    9270-0Glasgow coma score verbal7679923
    9268-4Glasgow coma score motor7676466
    38208-5Pain severity - Reported7569094
    50064-5Ideal body weight7297147
    9187-6Urine output6453291
    9269-2Glasgow coma score total5572112

    I have been mapping to the LOINC codes and came up short in blood pressure.  99% of blood pressures done worldwide are a pair of measurements made with a blood pressure cuff on a peripheral artery and 8480-6 mapped nicely to 72313002|Systolic arterial pressure|, I was good with the definition which CHARACTERIZES SYSTOLIC PHASE, PROPERTY PRESSURE, INHERES IN ARTERIAL STRUCTURE.  Systolic pressure measurements in general INHERE IN CARDIOVASCULAR STRUCTURE.  When I researched diastolic pressure, I found no observable for 'Diastolic arterial pressure' and all diastolic pressures have a consistent definition and INHERE IN CARDIOVASCULAR STRUCTURE.  Unfortunately, the fourth most common clinical observable does not exist in SNOMED CT but I would be willing to bet that we have LOTS of systems worldwide have been using 271650006|Diastolic blood pressure| to record that measurement over the years.   That would argue for changing the FSN of 271650006 and defining it properly but the best strategy alludes me.  Clearly the observables model supports proper definition of the LOINC concept but what should the map be in Nebraska Lexicon?

    Jim

  2. Given that the height (length) of the body is from sole to crown, would we define this as 

    ISA BODY HEIGHT MEASURE

    PROPERTY LENGTH

    SCALE TYPE QUANTITATIVE

    TIME ASPECT SINGLE POINT IN TIME

    GROUP 1 INHERES IN CROWN OF HEAD(SURFACE REGION)

    GROUP 2 SOLE OF FOOT(SURFACE REGION)?

  3. We are working on a computable phenotype for Alzheimers disease.  The pathology observables are not mysterious although there are a lot of new attribute/value pairs we need to add.  The clinical portion of the dataset will be heavily oriented towards neurocognitive examinations.  Building out the scoring of the Folstein minimental status exam, I am seeking some advice:

    72106-8 Minimental Status examination of cognitive function - total score(observable entity)

    ISA =311465003|Cognitive functions(observable entity)|

    INHERES IN =Cerebral structure

    TECHNIQUE =Assessment scale\NEW Folstein minimental status examination

    HAS REALIZATION = 719982003 Process\NEW Mental process(qualifier) \NEW |Cognition(qualifier)|

    SCALE =Quantitative

    TIME ASPECT =Single point in time

  4. Just one quick suggestion would be a PROPERTY = Ability or Function or similar, given that this is an observation to assess cognitive ability or function. Sometimes these tests are for finding presence of disease rather than normal function, not sure about this one.

    Further, whether this INHERES IN = CNS is also something to consider.



  5. I was looking for a template for the FUNCTION observables but could not find one.  Please give me a pointer if it exists.  I agree that INHERES IN is not an intuitive choice here, but arguably, if you remove the cerebrum, you certainly ablate cognition.  Frontal lobes and parieto-temporal and occipital regions are the most important in the case of cognitive function.

    Jim

  6. I want to model a couple observables for results of assessment scales for Pain severity and Chest pain.  Has there been preparatory work in what Observables flavor to use?

    Jim

  7. I have appended to the last meeting page the presentation from today to the Anesthesiology CRG on Glasgow Coma Scale and machine learning using the neurological exam