What is usage of “min_levels_of_separation” and “max_levels_of_separation”. They are two attributes defined in CONCEPT_ANCESTOR table? An example will be helpful.
Hi @pareshsinha,
Level of separation is a number of concepts between some ancestor_concept and descendant one in the hierarchy.
For each concept there can be several pathways to reach ancestor concept.
For example if you look for relationship between concept_id_1=4047584 (Lengthening of extraocular muscle)
and concept_id_2=4301351(Surgical procedure) you will see nothing. But you can find that:
4047584 (Lengthening of extraocular muscle) ***Is a*** 4214521 (Strabismus surgery) which in its turn ***Is a***
4132804 (Surgical procedure for clinical finding and/or disorder) which ***Is a*** 4301351(Surgical procedure)
Going these steps we see,that there are 3 relationships steps between
Lengthening of extraocular muscle and Surgical procedure
But there is another way:
4047584 (Lengthening of extraocular muscle) ***Is a*** 4214521 (Strabismus surgery) ***Is a***
4070457(Operation on extraocular muscle) ***Is a*** 4333146(Surgical procedure on orbit) ***Is a***
4154279(Surgical procedure on eye region) ***Is a*** 4233946(Operative procedure on head) ***Is a***
4301351(Surgical procedure)
Having the same start- and endpoints we have passed through 6 relationships.
For each pair of concepts (which are indirectly related with the series of intermediate concepts) “min_levels_of_separation” and “max_levels_of_separation” show resp. minimal and maximal number of
intermediate concepts between them.
Note: each concept refers to itself in CONCEPT_ANCESTOR table(for such case both “min_levels_of_separation” and “max_levels_of_separation” are equal to 0)
2 Likes
Thank you Eldar! Very helpful description.
Perhaps this answer can be added to the wiki?
http://www.ohdsi.org/web/wiki/doku.php?id=documentation:cdm:concept_ancestor
1 Like
A PR to this repository https://github.com/OHDSI/CommonDataModel/ with your write up please!
Concept ancestor is not clear to most of us.
Added with a slightly simpler example 
1 Like