Open mgurley opened 4 years ago
@mgurley can you please elaborate? Do you mean schedule components are administered?
@jeremywarner Can you elaborate what kind of temporal data points are available within the Hemonc.org ontology that we are not currently ingesting into OMOP? My naive guess would be 'number of cycles' and 'length of cycles'. I think the idea is that having expectations of the timing of oncology drug regimens will aide in their derivation.
Hi @mgurley my understanding of HemOnc temporality in this context would be the declarative relationships between two regimens which can take two flavors:
The reason that these classes are imbalanced is that many of the “can be preceded by” relationships are to surgical procedures, which don’t exist in a reciprocal format (e.g., AC can be proceeded by breast cancer surgery). All regimens should have a round trip so that if A is followed by B, B is preceded by A. It’s not perfect but it’s close.
The hang-up previously and the reason that Dima didn’t ingest is the “can be” – he didn’t like the optionality. In fact this is a bit of a placeholder because in reality the relationship is instantiated as part of a protocol, which is defined by a study design. A more expanded description would be:
Since the devil is in the details, what are the criteria? Generally speaking, the criteria are no obvious disease progression after A, and no undue toxicity that would prevent proceeding to B. So we’re back to treatment episodes…beyond the general there are some protocols that have very specific criteria which must be met before going from A to B, and there are other so-called response-adapted protocols that will determine whether one proceeds to B or to C or to something entirely different based on the response to treatment.
@jeremywarner @dimshitc @cgreich Actually, I think the temporality that we are wanting to next incorporate into the OMOP vocabulary is not the temporal relationships of the regimens to prior or subsequent therapies but, rather, the temporal properties of regimens in themselves. For example, let's look at the Hemonc.org page for the 'DE' regimen:
https://hemonc.org/wiki/Breast_cancer#DE
On this page it states:
For each regimen or regimen variant, I think we are wanting to be placed in the OMOP vocabulary the following properties
Gents: You have both cases we need to tackle.
@jeremywarner: I wouldn't worry about coding the condition under which a treatment can be followed by another one. That is up to the physician at treatment time. We are in the observational world, we just capture did it happen or not. So, I agree with @dimshitc that we should have proper bidirectional links, and that's it.
@mgurley: Jeremy has those timings, at least partially. We need to do two things: Put them into the vocabulary (currently not possible easily) and have the algorithm people start utilizing them.
This task is essential for the completion of the Milestone Rerun symposium plots
@sratwani I think you mean
this task is essential for the completion of Running an 'Oncology Drug Regimen Extraction Challenge’
Moving to backlog based on discussion with Jeremy on 12/13.
@mgurley can you please elaborate? Do you mean schedule components are administered?