| Abstract |
In this talk I will present some preliminary work that investigates the suitability of probabilistic graphical models to estimate parameters for a Dynamical System model of human functioning.
In the context an Ambient Intelligent System supporting depression therapy, Dynamic Bayesian Networks are proposed as a technique to model a patient's state from a set of measurements and observations recorded over time. Such a model will be exploited to relate temporal data to cognitive states whose levels are used as input parameters for a computational model of depression.
The presentation will not be about technical results, but instead about the challenges and motivations of such an approach to integrate data in model based reasoning.
|