regarding "CONCEPT_PATH and CONCEPT_CD do not align as depicted in this slide."
Well, yes, they do, the way we use i2b2 in HERON, at least. Again, I'm not sure how well I can explain this in email; we may need hackathon in-person discussion for it. But here goes... In HERON, we use a late-binding approach to connect concept_cd to concept_path. In the concept_cd, we just copy the native coding system from the EMR; i.e. epic DX_ID, COMPONENT_ID, MEDICATION_ID, etc. Then, in the concept_dimension and metadata tables, we relate, for example, the MEDICATION_ID for ANTABUSE 250 MG from the KUH formulary to the Disulfiram Oral Tablet SCDF from RxNorm (using gcn sequence number and/or NDC code as a cross-walk). We have a slide deck from a MedMapping<https://informatics.kumc.edu/work/wiki/MedMapping> design meeting that might help explain some of it: * Clarity Medication Mapping.pptx<https://informatics.kumc.edu/work/attachment/ticket/1246/Clarity%20Medication%20Mapping.pptx> [Download] <https://informatics.kumc.edu/work/raw-attachment/ticket/1246/Clarity%20Medication%20Mapping.pptx> (340.4 KB) Especially the "Resulting I2B2 Hierarchy" slide. Unfortunately, our DiagonsisMapping<https://informatics.kumc.edu/work/wiki/DiagnosisMapping> notes are even messier, though. And I wish the code were cleaner too... epic_dx_concepts.sql<https://informatics.kumc.edu/work/browser/heron_load/epic_dx_concepts.sql> Anyway... the point is: we can relate the native concept_cd's to any number of concept_path term hierarchies in a late-binding fashion; we can even have multiple such term hierarchies in the system at the same time; e.g. ICD9, ICD10, and snomed CT, all sharing the same facts. As to how we calculate, INSTANCE_NUM... it's somewhat involved... but that's more related to modifiers than concept_cd and concept_path, so more on that separately. -- Dan ________________________________ From: Greater Plains Collaborative Software Development [[email protected]] on behalf of Wilson Nathan [[email protected]] Sent: Wednesday, January 22, 2014 11:11 AM To: [email protected] Subject: PCORI Information Model Review -- Nathan Wilson (UW Madison) [...] Slide 10: CONCEPT_PATH and CONCEPT_CD do not align as depicted in this slide. * Multiple CONCEPT_PATHS to one CONCEPT_CD should look something like this. Only the SNOMED term should have multiple CONCEPT_PATHS because of poly-hierarchy CONCEPT_PATH CONCEPT_CD Diagnosis\ICD09\Diseases and Injuries\Diseases of the Digestive System\Appendicitis\Acute Appendicitis\Acute Appendicitis with Peritoneal Abscess DX|ICD09:540.1 Diagnosis\ICD10\Diseases of the Digestive System\Diseases of Appendix\Acute Appendicitis\Acute Appendicitis with Localized Peritonitis DX|ICD10:K35.3 Diagnosis\SNOMED\Clinical Finding\Disease\Inflammatory Disorder\Inflammation of Specific Body Structures or Tissue\Abscess of Trunk\Abdominal Abscess\Abscess of Peritoneum\Acute Appendicitis with Peritoneal Abscess DX|SNOMED:51036000 Diagnosis\SNOMED\Clinical Finding\Finding by Site\Disorder by Body Site\Disorder of Soft Tissue\Disorder of Soft Tissue of Trunk\Acute Appendicitis with Peritoneal Abscess DX|SNOMED:51036000 * Because of how i2b2 queries its data, we cannot say that each of the following concepts have the same path unless each concepts ancestors and dependents also have a 1-to-1 mapping. Slide 12: How are we calculating the value(s) for INSTANCE_NUM? Nathan S. Wilson Ontologist Health Information Management Center (HIMC) ______________________________________________________________________________________________________ 301 S. Westfield Rd - Suite 320 Madison, WI 53717 Phone: (608) 890-5780 eMail: [email protected]
