harvest-vocab

harvest-vocab is a Harvest Stack app provides abstract models for defining vocabulary-like models and building a corresponding index for hierarchical self-related data.

For example, this is how you could define models for storing ICD9 codes:

from vocab.models import AbstractItem, AbstractItemIndex

class Diagnosis(AbstractItem):
 description = models.CharField(max_length=50)
 code = models.CharField(max_length=10)
 parent = models.ForeignKey('self', related_name='children')

ICD9 codes are hierachical therefore when I ask the questions, "Give me all the patients who have a diagnosis in ICD9 367 (Disorders of refraction and accommodation)", then this should not only query 367, but all descendent diagnoses as well (which includes another 2 levels).

This kind of query becomes difficult to write since you only have access to the direct parent of the a particular diagnosis, thus the query would look like this.

from django.db.models import Q
Diagnosis.objects.filter(Q(code='367') | Q(parent__code='367'))

The obvious problem here is that any diagnoses 2+ levels down from '367' are not included.

Create A Flat Index

To alleviate this issue, an AbstractItemIndex subclass can be defined which will build a flat index for an AbstractItem subclass. Simply define it like this:

class DiagnosisIndex(AbstractItemIndex):
 item = models.ForeignKey(Diagnosis, related_name='item_indexes')
 parent = models.ForeignKey(Diagnosis, related_name='parent_indexes')

# builds the index for Diagnosis
DiagnosisIndex.objects.index()

The last line generates a flat index of the hierarchy which alleviates the unknown depth issue. So now, the same question stated above can be answered this way:

# either the item has this code or one of it's parents has this code
condition = Q(item__code='367') | Q(parent__code='367')
item_ids = DiagnosisIndex.objects.filter(condition).values_list('item__id', flat=True)
diagnoses = Diagnosis.objects.filter(id__in=item_ids)

Requirements:

• Python