OHDSI Home | Forums | Wiki | Github

How to represent family history

Vocabulary Users
#61

What if you split up the periods into the 1-day facts and make it this way:

For the patient who had a fact of contact with another infected person, create the observation records:
observation_date = a date of 1-day fact.
observation_id = 37311059 (Exposure to SARS-CoV-2) or any of its children.

Then connect the exposure facts with the objects the exposure happened to through the fact_relationship table:
domain_concept_id_1 = 1147304 (Observation table).
fact_id_1 = id of the exposure fact.
domain_concept_id_2 = 1147314 (Person table).
fact_id_2 = id of the person the exposure happened to.
relationship_concept_id = there’s no good one, but custom “Event to Subject” is something you can start with.

Unfortunately, the analysis will be painful. To go from one individual to another the path would be:
PERSON - OBSERVATION - FACT_RELATIONSHIP - PERSON tables.

1 Like
#62

good suggestion. But I also need to relate one person to another, one person who was infected with COVID-19, and the other person who was exposed to that other infected person.

#63

this is what Alex suggests, you relate one person to another through OBSERVATION and FACT_RELATIONSHIP.
And you can’t do just PERSON - FACT_RELATIONSHIP - PERSON because you need to capture the exact period and also Observation table will have the exact event = 37311059 (Exposure to SARS-CoV-2)

1 Like
#64

Ok, I got it, thanks.

#65

Dear Community,

As the overhaul of SNOMED is in progress, it’s a good time to revive the discussion concerning the hierarchy of Family History.

As you can see, the SNOMED hierarchy of Family History concepts is quite messy. We propose the creation of a specific hierarchy of new concepts:

  1. Our hierarchy could follow the structure of a genealogical tree with several orders of ancestry and include information about maternal and paternal relatives. If the information is not available, it probably requires another branch.
  2. We can omit the information about non-blood relatives, such as spouses, step-parents, step-children, step-siblings, etc. Or is there a use case?
  3. We can create pre-coordinated concepts that carry time-related information (e.g., decades of life) and embed them into the hierarchy of relatives. I would be happy to learn what time-related information is relevant to your research. And we can also discuss what other facts should be pre-coordinated in our hierarchy.
  4. We also should discuss the effort and time needed to create the above-mentioned hierarchy and to perform respective mappings.

We’ll be happy to hear your thoughts and ideas about this approach to the hierarchy of Family History.

Also, I invite everyone to discuss these questions in the scope of the WG call dedicated to SNOMED overhaul.

Regards,

Masha and the Vocabulary Team.

1 Like
t