This shows you the differences between two versions of the page.
Next revision | Previous revision | ||
documentation:oncology:meeting_notes_2019_aug-27 [2019/08/28 08:56] mgurley created |
documentation:oncology:meeting_notes_2019_aug-27 [2019/08/28 14:51] (current) mgurley |
||
---|---|---|---|
Line 1: | Line 1: | ||
- | **Meeting Notes August 13, 2019** | + | **Meeting Notes August 27, 2019** |
- | + | ||
- | Here is a link to a recording of the meeting: [[https://bluejeans.com/s/VUkqB|recording]]. | + | |
**Topics** | **Topics** | ||
- | * Robert Miller shared and explained his code to convert NAACCR flat file-formatted data to the input format necessary for the generic ETL SQL. See [[https://github.com/OHDSI/OncologyWG/blob/master/documentation/OHDSI%20Oncology%20Module%20-%20Crosswal%20from%20Cancer%20Registry.pdf | + | |
- | |here]] generic ETL SQL . | + | * Seng Chan You from Ajou University told the group about the code he and his colleagues have developed for extracting treatment cycle from single medication records in CDM databases from OMOP data. See here: https://github.com/ABMI/treatmentCycleExtraction |
- | * Michael Gurley shared and explained his code to convert the data model from the popular tumor registry software Elekta METRIQ to the input format necessary for the generic ETL SQL. | + | * We reviewed the vocabulary tasks from the workshop. |
- | * Andrew Williams raised the issue whether the two current NAACCR input formats sufficiently cover how tumor registry data will be encountered in the wild. It was countered that the best we can do is harvest efforts as new adopters encounter new scenarios. | + | * Add the EPISODE table and field concepts to the vocabulary for metadata field reference. Like ‘episode.episode_id’. See here: https://github.com/OHDSI/OncologyWG/issues/66 |
- | * Andrew Williams and Maxim Moinat raised the issue of whether folks coming from non-NAACCR sources should map their tumor characteristics to the NAACCR vocabulary. It was argued folks should map to NAACCR so we can make tumor registry data and non-tumor registry data comparable. It was pointed out the problem arises that just saying that we should pick the "standard" concept for each tumor characteristic runs into the "Choose Your Own Adventure" problem that OMOP contains multiple "standard" concepts for many tumor characteristics semantic concepts. The approach we have agreed as a group to take is to choose NAACCR as our standard for tumor characteristics for the time being until we can map NAACCR to a more proper standardized vocabulary like SNOMED or LOINC. Hopefully via the Nebraska Lexicon. See the {{:documentation:oncology:ohdsi_oncology_meeting_06_18_2019.pdf| Presentation}} from June 18 where this issue was discussed. It sounds like this issue is still open for debate. An alternative to using NAACCR as the standardized tumor characteristics vocabulary is to let people use whatever seems right and depend on the mapping between concepts being worked out at the analysis stage. This issue can be discussed at the hackathon. | + | * Create new 'Regimen' domain, move all the Hemonc.org regimens into the new domain, create entries in the new domain for NAACCR high-level drug concepts and non-standardize existing drug NAACCR item codes. See here:https://github.com/OHDSI/OncologyWG/issues/69. |
- | * The two-day OHDSI Oncology hackathon on August 21st and 22nd was re-announced. Details will be emailed to all members of the OHDIS Oncology Workgorup. | + | * Put ‘did not happen’ NAACCR concepts in the Observation domain. For example, ‘Patient did not have chemotherapy’. See here: https://github.com/OHDSI/OncologyWG/issues/68 |
+ | * Fix missing leading zeros in the concept_codes for Treatment NAACCR item codes. See here: https://github.com/OHDSI/OncologyWG/issues/60 | ||
+ | * The 'NAACCR Proc Schema' entries do not have all the necessary relationships to ICDO codes via 'Proc Schema to ICDO' See here: https://github.com/OHDSI/OncologyWG/issues/60 | ||
+ | * We made a decision on a convention for making a non-ambiguous reference between a 'Disease Episode' and 'Treatment Episode'. Convention: Populate parent_id of a ‘Treatment Episode’ with the ‘Disease Episode’ it is 'treating'. | ||
+ | * For the symposium, we decided to skip all the ICDO codes that participate in multiple NAACCR schemas. Until we can create a mapping table to be used for handling ambiguous ICDO site/histology combinations present in multiple schemas. See here: https://github.com/OHDSI/OncologyWG/issues/59. See here for SQL to identify the ambiguous ICDO Site/Histology combinations: https://github.com/OHDSI/OncologyWG/blob/master/documentation/debug_naaccr_ambiguites.sql | ||
+ | |||