Hi! i have to model a problem: in a hospital 150 patients have done 200 different exercises. Each exercise assess one cognitive function of the patient (Attention, Memory, Planning,..) when the patient finishes de exercise he gets the result a number from 0 (if he did everything wront) to 100 if he did everything OK.
Each patient did exercises during around 5 months, each exercise could be done by each patient more than once, so now i have a database of 40000 exercises done by those 150 patients. The number of exercises done by each patient is different for all of them, in the database i have just the results, the name of each exercise, the date of execution and the patient identification.
The cognitive state of each patient is evaluated using a 10 questions test before and after doing all the exercises so I also have the results of all this evaluations, each patient gets a 0 to 4 punctuation in each of the 10 questions of this test. I mean that the patient is evaluated once with this 10 questions, then does all the exercises during around 5 months and then he is evaluated again with the same 5 questions.
In the data base there's also for each patient the results of these 2 evaluations.
Now given any patient A and one of the exercises E at any moment T during the 5 months that they are doing the exercises I need to find similar patients to A who had done the exercise E in a similar moment T and return the results that those similar patients got in E.
Sorry for the so long explanation, I think that this could be modeled using a dynamic bayesian network and maybe someone could tell me his opinion... I'm new to bayesian networks and need some hints on this..