baukearends
commented
4 weeks ago @hcadavid Difficult case... Setting Problems for all conditions that are confirmed absent would lead to a wide array of Problem resources for each participant, and will definitely not resemble any PGO data I presume. However, I do understand the need for a distinction between confirmed negatives and missing data.
Do you currently use the VerificationStatusCodelist for Problems? If constructing a Problem for all negative cases, assigning a 'Known absent' verification status might be the best approach.
hcadavid
Based on the pairing rules defined, there are two possible outcomes when generating the FHIR resources that represent a condition (e.g., diabetes, MI, HF, etc): the FHIR resource is created, with an 'Active' status (when it is reported as such), or it is just not created otherwise, as there is no such thing as 'Inactive' diabetes, HF, MI, etc.
https://github.com/MyDigiTwinNL/CDF2Medmij-Mapping-tool/blob/071cb66c2fa3a625d813e082556860a74b6b63f4/src/lifelines/Diabetes.ts#L46-L57
However, this causes a limitation on the generated data, as there is no way to tell which participants don't have a FHIR Problem/Condition because they explicitly say 'NO' on the related questions, and which don't have it because they didn't answer the related questions (missing data). As there are only two accepted outcomes for the Problem resource the only solution I see for this is to also generate the Problem/FHIR resource for participants that explicitly reported that never had the condition, but setting it as an 'Inactive' one. With this, we will be able to identify three sets for each condition: the ones with the condition, the ones that never had it, and the ones for whom there is no certainty. Knowing the latter would also help when doing data imputation, if needed.
@baukearends could you give me your thoughts on this?