How to infer the "Characteristic" Properties

classic Classic list List threaded Threaded
6 messages Options
Reply | Threaded
Open this post in threaded view
|

How to infer the "Characteristic" Properties

SAPI
Hello there,

I am doing a disease diagnosis system, I know how to make a class with N&S condtions so that inferrence engine will tell you what is the disease if I enter all the symptoms under the hasSymtoms property of the disease class. Now say I have three different diseases D1, D2 & D3 each has different symptoms has shown below:
D1: S1, S2, S3, S4,
D2: S2, S3, S4, S5
D3: S3, S4, S5, S6, S7

If the patient tell you he hasSymptoms S3 & S4, as these are the common symptoms across D1..D3, then these are useless, but if they tell you he has S1, then you know he has D1 immediately, if he has S6 & S7, then he has D3. So is there a way to infer the characteristic property of D1..D3 as follows:
D1: S1
D2: S2
D3: S6, S7

Also is there any way to display a list of all possible symptoms, ie S1,S2,S3..S7?

Any feedback will be highly appreciated.

SAPI
Reply | Threaded
Open this post in threaded view
|

Re: How to infer the "Characteristic" Properties

Bill Dickinson
I can't speak for the folks who write the reasoners, but, without intending
to complicate this scenario, I wonder if you might also need to include a
time of sympton onset.  Its seems reasonable that D1, D2, and D3 are all
valid considerations for S3 which is shared by all three.  If S2 shows up
later, re-eval would strongly infer D2, but wouldn't all three diseases be
candidates if only S3 was presenting?   What if only S3 and S4 were present?
(same conclusion).

Now S5 shows up - then we can rule out D1, I think, but can't decide if D2
or D3 (or both) is correct until S2 shows up, or s7.  Maybe there is no time
involved.  Maybe the syptoms all show up at once - I guess it depends on the
disease.  Oh, well, just some random thoughts.  I think that inferences will
change as the symptoms appear.  I understand you wish to define a
"characteristic" of each disease based on the unique sysmptoms of each, but
I think your "patient" is not going to present with, say, only S1, or only
s2, or only S6 and S7 when there are other sysmptoms present AND some of
them are shared by all diseases.

Regards,

Bill Dickinson
----- Original Message -----
From: "SAPI" <[hidden email]>
To: <[hidden email]>
Sent: Saturday, January 19, 2008 10:04 PM
Subject: [protege-owl] How to infer the "Characteristic" Properties


>
> Hello there,
>
> I am doing a disease diagnosis system, I know how to make a class with N&S
> condtions so that inferrence engine will tell you what is the disease if I
> enter all the symptoms under the hasSymtoms property of the disease class.
> Now say I have three different diseases D1, D2 & D3 each has different
> symptoms has shown below:
> D1: S1, S2, S3, S4,
> D2: S2, S3, S4, S5
> D3: S3, S4, S5, S6, S7
>
> If the patient tell you he hasSymptoms S3 & S4, as these are the common
> symptoms across D1..D3, then these are useless, but if they tell you he
> has
> S1, then you know he has D1 immediately, if he has S6 & S7, then he has
> D3.
> So is there a way to infer the characteristic property of D1..D3 as
> follows:
> D1: S1
> D2: S2
> D3: S6, S7
>
> Also is there any way to display a list of all possible symptoms, ie
> S1,S2,S3..S7?
>
> Any feedback will be highly appreciated.
>
> SAPI
> --
> View this message in context:
> http://www.nabble.com/How-to-infer-the-%22Characteristic%22-Properties-tp14978328p14978328.html
> Sent from the Protege - OWL mailing list archive at Nabble.com.
>
> _______________________________________________
> protege-owl mailing list
> [hidden email]
> https://mailman.stanford.edu/mailman/listinfo/protege-owl
>
> Instructions for unsubscribing:
> http://protege.stanford.edu/doc/faq.html#01a.03
>

_______________________________________________
protege-owl mailing list
[hidden email]
https://mailman.stanford.edu/mailman/listinfo/protege-owl

Instructions for unsubscribing: http://protege.stanford.edu/doc/faq.html#01a.03 
Reply | Threaded
Open this post in threaded view
|

Re: How to infer the "Characteristic" Properties

Alan Ruttenberg-2
In reply to this post by SAPI


On Jan 19, 2008, at 11:04 PM, SAPI wrote:

>
> Hello there,
>
> I am doing a disease diagnosis system, I know how to make a class  
> with N&S
> condtions so that inferrence engine will tell you what is the  
> disease if I
> enter all the symptoms under the hasSymtoms property of the disease  
> class.
> Now say I have three different diseases D1, D2 & D3 each has different
> symptoms has shown below:
> D1: S1, S2, S3, S4,
> D2: S2, S3, S4, S5
> D3: S3, S4, S5, S6, S7
>
> If the patient tell you he hasSymptoms S3 & S4, as these are the  
> common
> symptoms across D1..D3, then these are useless, but if they tell you  
> he has
> S1, then you know he has D1 immediately, if he has S6 & S7, then he  
> has D3.
> So is there a way to infer the characteristic property of D1..D3 as  
> follows:
> D1: S1
> D2: S2
> D3: S6, S7

If you have symptomOf (inverse of hasSymptom) then you could ask for  
something like:

symptomOf D1 and not (symptomOf D2) and not (symptomOf D3) ...


> Also is there any way to display a list of all possible symptoms, ie
> S1,S2,S3..S7?

the subclasses of symptomOf some Disease

Though I would expect you would have a commons superclass for symptoms.

You might aim general OWL questions to [hidden email]

Biomedical related questions might also get some response from [hidden email]

If you are interested in this sort of thing, you could also review

http://lists.w3.org/Archives/Public/public-semweb-lifesci/2007Feb/0076.html

Regards,
Alan

>
>
> Any feedback will be highly appreciated.
>
> SAPI
> --
> View this message in context: http://www.nabble.com/How-to-infer-the-%22Characteristic%22-Properties-tp14978328p14978328.html
> Sent from the Protege - OWL mailing list archive at Nabble.com.
>
> _______________________________________________
> protege-owl mailing list
> [hidden email]
> https://mailman.stanford.edu/mailman/listinfo/protege-owl
>
> Instructions for unsubscribing: http://protege.stanford.edu/doc/faq.html#01a.03

_______________________________________________
protege-owl mailing list
[hidden email]
https://mailman.stanford.edu/mailman/listinfo/protege-owl

Instructions for unsubscribing: http://protege.stanford.edu/doc/faq.html#01a.03 
Reply | Threaded
Open this post in threaded view
|

Re: How to infer the "Characteristic" Properties

nassim nassim
for that ...   maybe  you have  to use SWRL  and a rule engine  
(algernon,sweetrule,jess ....) ..

ps: sorry for my bad english



>From: Alan Ruttenberg <[hidden email]>
>Reply-To: User support for the Protege-OWL
>editor<[hidden email]>
>To: User support for the Protege-OWL editor
><[hidden email]>, SAPI <[hidden email]>
>Subject: Re: [protege-owl] How to infer the "Characteristic" Properties
>Date: Sun, 20 Jan 2008 02:25:04 -0500
>
>
>
>On Jan 19, 2008, at 11:04 PM, SAPI wrote:
>
> >
> > Hello there,
> >
> > I am doing a disease diagnosis system, I know how to make a class
> > with N&S
> > condtions so that inferrence engine will tell you what is the
> > disease if I
> > enter all the symptoms under the hasSymtoms property of the disease
> > class.
> > Now say I have three different diseases D1, D2 & D3 each has different
> > symptoms has shown below:
> > D1: S1, S2, S3, S4,
> > D2: S2, S3, S4, S5
> > D3: S3, S4, S5, S6, S7
> >
> > If the patient tell you he hasSymptoms S3 & S4, as these are the
> > common
> > symptoms across D1..D3, then these are useless, but if they tell you
> > he has
> > S1, then you know he has D1 immediately, if he has S6 & S7, then he
> > has D3.
> > So is there a way to infer the characteristic property of D1..D3 as
> > follows:
> > D1: S1
> > D2: S2
> > D3: S6, S7
>
>If you have symptomOf (inverse of hasSymptom) then you could ask for
>something like:
>
>symptomOf D1 and not (symptomOf D2) and not (symptomOf D3) ...
>
>
> > Also is there any way to display a list of all possible symptoms, ie
> > S1,S2,S3..S7?
>
>the subclasses of symptomOf some Disease
>
>Though I would expect you would have a commons superclass for symptoms.
>
>You might aim general OWL questions to [hidden email]
>
>Biomedical related questions might also get some response from
>[hidden email]
>
>If you are interested in this sort of thing, you could also review
>
>http://lists.w3.org/Archives/Public/public-semweb-lifesci/2007Feb/0076.html
>
>Regards,
>Alan
>
> >
> >
> > Any feedback will be highly appreciated.
> >
> > SAPI
> > --
> > View this message in context:
>http://www.nabble.com/How-to-infer-the-%22Characteristic%22-Properties-tp14978328p14978328.html
> > Sent from the Protege - OWL mailing list archive at Nabble.com.
> >
> > _______________________________________________
> > protege-owl mailing list
> > [hidden email]
> > https://mailman.stanford.edu/mailman/listinfo/protege-owl
> >
> > Instructions for unsubscribing:
>http://protege.stanford.edu/doc/faq.html#01a.03
>
>_______________________________________________
>protege-owl mailing list
>[hidden email]
>https://mailman.stanford.edu/mailman/listinfo/protege-owl
>
>Instructions for unsubscribing:
>http://protege.stanford.edu/doc/faq.html#01a.03

_________________________________________________________________
MSN Messenger: appels gratuits de PC à PC !
http://www.msn.fr/msger/default.asp

_______________________________________________
protege-owl mailing list
[hidden email]
https://mailman.stanford.edu/mailman/listinfo/protege-owl

Instructions for unsubscribing: http://protege.stanford.edu/doc/faq.html#01a.03 
Reply | Threaded
Open this post in threaded view
|

Re: How to infer the "Characteristic" Properties

kamalej seapho
In reply to this post by SAPI
you can use UNION concept for this case and the application them've ADD  Property for Add-in eg. Just create S4 in D1 and Add into  D2 and D3 for reuse.

SAPI <[hidden email]> wrote:

Hello there,

I am doing a disease diagnosis system, I know how to make a class with N&S
condtions so that inferrence engine will tell you what is the disease if I
enter all the symptoms under the hasSymtoms property of the disease class.
Now say I have three different diseases D1, D2 & D3 each has different
symptoms has shown below:
D1: S1, S2, S3, S4,
D2: S2, S3, S4, S5
D3: S3, S4, S5, S6, S7

If the patient tell you he hasSymptoms S3 & S4, as these are the common
symptoms across D1..D3, then these are useless, but if they tell you he has
S1, then you know he has D1 immediately, if he has S6 & S7, then he has D3.
So is there a way to infer the characteristic property of D1..D3 as follows:
D1: S1
D2: S2
D3: S6, S7

Also is there any way to display a list of all possible symptoms, ie
S1,S2,S3..S7?

Any feedback will be highly appreciated.

SAPI
--
View this message in context: http://www.nabble.com/How-to-infer-the-%22Characteristic%22-Properties-tp14978328p14978328.html
Sent from the Protege - OWL mailing list archive at Nabble.com.

_______________________________________________
protege-owl mailing list
[hidden email]
https://mailman.stanford.edu/mailman/listinfo/protege-owl

Instructions for unsubscribing: http://protege.stanford.edu/doc/faq.html#01a.03


Looking for last minute shopping deals? Find them fast with Yahoo! Search.
_______________________________________________
protege-owl mailing list
[hidden email]
https://mailman.stanford.edu/mailman/listinfo/protege-owl

Instructions for unsubscribing: http://protege.stanford.edu/doc/faq.html#01a.03 
Reply | Threaded
Open this post in threaded view
|

Re: How to infer the "Characteristic" Properties

SAPI
In reply to this post by Bill Dickinson
Thank you all for all the feedback.. My original questions has been over simplified a little bit...
The D1,D2,D3 are in fact disease syndromes say DS1 DS2 DS3 which fall under a single disease, think of the disease 'Flu' has multiple syndromes, so the fact that S3 and S4 tell us she has a flu does not help much (the doctor would have known in most cases after patient speaks of all her symptoms), the doctor has to asked if she has S1 or S2 etc so that he can prescribe according to the syn DS1 or DS2... Thats why it is important that he can retrieve the characteristics symptoms and all the symptoms of flu for him to check with the patient. Yes, the hasOnset plus some more is also taken into account in the current ontology, I do not want to bring them in to complicate the problems here.

I can hardcode a class with the N&S conditions like Alan described, e.g. symptomOf D1 and not (symptomOf D2) and not (symptomOf D3) ... to do the filtering, but this might mean we have to maintain the conditions as D4, D5 .. are added, this amount to maintaining a few thousand of classes in the current ontology. Hopefully it can be just a single icon or it is a function you can select and execute (like "computer types"), this might required some coding to the basic reasoner engine, I am not sure.

-Sapi

Bill Dickinson wrote
I can't speak for the folks who write the reasoners, but, without intending
to complicate this scenario, I wonder if you might also need to include a
time of sympton onset.  Its seems reasonable that D1, D2, and D3 are all
valid considerations for S3 which is shared by all three.  If S2 shows up
later, re-eval would strongly infer D2, but wouldn't all three diseases be
candidates if only S3 was presenting?   What if only S3 and S4 were present?
(same conclusion).

Now S5 shows up - then we can rule out D1, I think, but can't decide if D2
or D3 (or both) is correct until S2 shows up, or s7.  Maybe there is no time
involved.  Maybe the syptoms all show up at once - I guess it depends on the
disease.  Oh, well, just some random thoughts.  I think that inferences will
change as the symptoms appear.  I understand you wish to define a
"characteristic" of each disease based on the unique sysmptoms of each, but
I think your "patient" is not going to present with, say, only S1, or only
s2, or only S6 and S7 when there are other sysmptoms present AND some of
them are shared by all diseases.

Regards,

Bill Dickinson
----- Original Message -----
From: "SAPI" <m59sapi@yahoo.com>
To: <protege-owl@lists.stanford.edu>
Sent: Saturday, January 19, 2008 10:04 PM
Subject: [protege-owl] How to infer the "Characteristic" Properties


>
> Hello there,
>
> I am doing a disease diagnosis system, I know how to make a class with N&S
> condtions so that inferrence engine will tell you what is the disease if I
> enter all the symptoms under the hasSymtoms property of the disease class.
> Now say I have three different diseases D1, D2 & D3 each has different
> symptoms has shown below:
> D1: S1, S2, S3, S4,
> D2: S2, S3, S4, S5
> D3: S3, S4, S5, S6, S7
>
> If the patient tell you he hasSymptoms S3 & S4, as these are the common
> symptoms across D1..D3, then these are useless, but if they tell you he
> has
> S1, then you know he has D1 immediately, if he has S6 & S7, then he has
> D3.
> So is there a way to infer the characteristic property of D1..D3 as
> follows:
> D1: S1
> D2: S2
> D3: S6, S7
>
> Also is there any way to display a list of all possible symptoms, ie
> S1,S2,S3..S7?
>
> Any feedback will be highly appreciated.
>
> SAPI
> --
> View this message in context:
> http://www.nabble.com/How-to-infer-the-%22Characteristic%22-Properties-tp14978328p14978328.html
> Sent from the Protege - OWL mailing list archive at Nabble.com.
>
> _______________________________________________
> protege-owl mailing list
> protege-owl@lists.stanford.edu
> https://mailman.stanford.edu/mailman/listinfo/protege-owl
>
> Instructions for unsubscribing:
> http://protege.stanford.edu/doc/faq.html#01a.03
>

_______________________________________________
protege-owl mailing list
protege-owl@lists.stanford.edu
https://mailman.stanford.edu/mailman/listinfo/protege-owl

Instructions for unsubscribing: http://protege.stanford.edu/doc/faq.html#01a.03