[protege-discussion] Schizophrenic model?

[Samson Tu]

Hi,

Here is one approach that should work:

1. Allow a Finding hierarchy of any granularity, so you may have
    Delusions
       Bizarre delusion
       Non-Bizarre delusion
       Mood-congruent
       Mood-neutral
    Hallucinations
    Disorganized speech
    Grossly disorganized behavior
    Catatonic behavior)

2. Define a Schizophrenia-defining-criterion hierarchy with the 4 
subclasses Delusion_criterion, ... 
Grossly_disorganized_or_catatonic_behavior

3. Create singleton individuals for each Schizophrenia-defining-criterion

4. Define rules of the form:
    Person(?p)and has_symptom(?p, ?s) and Delusions(?s) => 
satisfy_criterion(?p, d), where d is the singleton individual of the 
Delusions_criterion class

5. Define Schizophrenia as a disease where a person's satisfy_criterion 
is min 2 Schizophrenia-defining-criterion

The point is that Schizophrenia definition counts the number of criteria 
that are satisfied, not the number of symptoms present. Symptoms may be 
very granule.

Another approach is to take combinations of 2 symptom classes from the 5 
in the list of criteria.


Samson

maarten wrote:
> Hi all,
> 
> I have the following problem modeling symptoms of schizophrenia. Someone 
> is diagnosed as schizophrenic when he/she (among other things) has at 
> least 2 of the following symptoms:
> 1. Delusions
> 2. Hallucinations
> 3. Disorganized speech
> 4. (Grossly disorganized behavior or Catatonic behavior)
> 
> The problem is that I'm unable to avoid that some is diagnosed as 
> schizophrenic if he/she has both symptoms 'Grossly disorganized 
> behavior' and 'Catatonic behavior' and no other symptoms, while they 
> should count for 1 only.
> 
> I have tried to model this in a straightforward way:
> - define a defined class GC consisting of Grossly .. or Catatonic behavior)
> - define an auxiliary SchizophreniaSymptom consisting of 1-3 and GC
> - define someone schizophrenia as someone with min 2 of symptoms of 
> SchizophreniaSymptom
> (see attachment)
> 
> but no success...
> 
> Is it possible to model this correctly, so that someone with both both 
> symptoms from 4 are not doubly counted? Any pointers in the right 
> direction appreciated. Thanks.
> 
> Kind regards,
> 
> Maarten