Hierarchy

Presentation

To generate it, it uses the fact_relationship OMOP table, with the care_site domain and the A is part of B relation. Thus, it generates a wide-type table, effectively flattening out the hierarchical structure of each care site.

This dataset is useful to find the parent of a care_site, e.g.:

• in which hospital is this UDS (Unité De Soin) ?
• in which UF (Unité Fonctionnelle) is this UMA (Unité Médico-Administrative) ?

Structure and usage

In this dataset each row corresponds to a given care_site and the columns contain the ids of the parent care_site for several hierarchical level. Those columns are thus values contained in care_site_type_source_value.

Internally, the dataset is returned by calling the function get_care_site_hierarchy():

from eds_scikit.resources import registry
df = registry.get("data", function_name="get_care_site_hierarchy")()

Use your own data.

It is as simple as registering a new loading function:

from eds_scikit.resources import registry # (1)

@registry.data("get_care_site_hierarchy") # (2)
def get_care_site_hierarchy():
      """
      Your code here
      """
      return df

1. The registry instance stores user-defined functions
2. Using this decorator allows to register the function when importing the corresponding file

Then simply import your custom_resources module before running eds-scikit's pipelines, and you're good to go.

Structure and usage

Internally, the dataset is returned by calling the function get_care_site_hierarchy(). It should return a Pandas Dataframe with the following columns:

• care_site_id (OMOP column): The identifier of the care site
• care_site_type_source_value (OMOP column): The type of care site

Additionally, it can contains an arbitrary number of columns whose name are values from care_site_type_source_value, and whose values are care_site_id of the corresponding parent structure

Generation function

You can generate the dataset on your specific data using this function