Analysis design question...

Discussion in 'microsoft.public.sqlserver.datamining' started by VictorV, May 11, 2007.

  1. VictorV

    VictorV Guest

    Hi.

    I have a little design question.

    I work in a healthcare industry. In one of our table we have evaluation info
    for each patient.
    There are 10 different score fields (with values 0,10,20,30...), then there
    is a Total_Score fields,
    and the Date of Evaluation fields, and the Type of Eval. The evaluations are
    done quarterly for each patient.
    So, the table might look like this:

    PatientID Score1 Score2 Score3 ... Score10 Total_Score Date_Of_Eval
    Type_Of_Eva
    --------------------------------------------------------------------------------
    1 0 10 0 30 40 01/01/2006 Initial
    1 0 20 0 20 40 04/01/2006 Quarterly
    1 20 20 0 10 50 07/01/2006 Discharge
    2 10 10 0 30 50 02/01/2006 Initial
    2 10 10 0 30 50 05/01/2006 Quarterly
    2 10 0 0 30 40 08/01/2006 Quarterly
    2 10 0 0 30 40 11/01/2006 Quarterly
    .......................

    So, as you can see, each patient has their own cycle of evaluations.
    Also, we have another table where we have the services (procedures) that are
    done to those patients.
    Each patient might have many different services done (identified by
    Procedure Codes) doring each month,
    or they might have some months without any services at all.

    So the simplified table might look like this:

    PatientID Date_of_Service Procedure_Code Units Cost
    ---------------------------------------------------
    1 01/01/2006 A0001 5 $50
    1 01/11/2006 A0002 2 $20
    1 02/01/2006 A0001 3 $30
    1 03/01/2006 A0005 4 $10
    1 05/01/2006 A0006 5 $50
    2 02/01/2006 A0007 1 $50
    2 03/01/2006 A0003 5 $20
    2 03/01/2006 A0002 4 $30
    2 04/01/2006 A0002 5 $40
    2 06/01/2006 A0002 3 $20
    2 07/01/2006 A0007 2 $50

    So, our analysts want to do some analysis on all that info.
    They want to be able to ask questions like:
    1. What services were provided for patients with the Score2 = 20 ?
    2. What services were done for patients for their Total_Score to improve
    (which mean the score become lower).
    3. What services were done for patients for their Score3 to get worse from
    the Initial Eval to the current state.

    And so on....

    So, my questions are: Is it possible to achieve this kind of analysis by
    creating some kind of a cube ?
    Or would I have to use a SQL Server 2005 Data Mining feature for that ? Or
    maybe some other statistical tools like SPSS ?
    What about SPSS Statistical Services for SQL Server 2005 ? Did anyone use it
    already ?

    Please, any input would be appreciated. This project is very urgent.

    Thank you,

    Victor.
     
    VictorV, May 11, 2007
    #1
    1. Advertisements

Ask a Question

Want to reply to this thread or ask your own question?

You'll need to choose a username for the site, which only take a couple of moments (here). After that, you can post your question and our members will help you out.