Re: [Gnumed-devel] Family History tables

[Karsten Hilbert]

On Mon, Jun 13, 2011 at 11:45:45AM -0700, Jim Busser wrote:

> - non-blood relatives may have significance in terms of
> transmissible diseases, however this is best modelled by
> identical matches on demographic address (geolocation)

This can be entered into "Family" History under an
appropriate (non-blood) relationship type.

> - genealogical relationships in GNUmed could be modelled
> using ideas from http://www.defoam.net/hubris/hubris03.htm
> (1997) which modelling seems to be built around persons who
> are *in* the database at the time

There is, currently, no intent to model genealogy. The
Hubris considerations are helpful in understanding the
issuse nonetheless.

> - and relationship records would be distinct from family
> history *items*

At this stage each FHx item contains an FK into the
relationship type table. Values provided by default are take
from Richard's documentation.

> each of which may (or may not) populate an
> f-key to an actual person in GNUmed

Not at this stage.

> (or woudl it be an f-key
> to a genealogical relationship record?) …

Again, currently no intent to support pedigrees.

> and a separate
> field which specifies the relationship (maternal grandmother
> etc) to the patient, which can be populated from patient
> self-report even when this grandmother was not created in
> GNUmed?

That's the above-mentioned FK into clin.fhx_relation_type.

> - what about when the patient just remembers "it was an
> uncle" but cannot remember maternal or paternal or at what
> age?

The GNUmed user selects "Relationship: uncle". The age-onset
would be left blank.

> - if we would build, for any patient, a list of
> genealogical relationships to particular people then those
> people should have to be created as people in GNUmed, yes?

Yes.

Karsten
-- 
GPG key ID E4071346 @ gpg-keyserver.de
E167 67FD A291 2BEA 73BD  4537 78B9 A9F9 E407 1346