AN ONLINE TRAINING ASSESSMENT BASED ON NAIVE BAYES NETWORK MODELED BY GAUSSIAN MIXTURE MODELS FOR MEDICAL SIMULATORS

Abstract

Gaussian Mixture Models has been used inseveral areas to model complex statistical distribution,mainly for decision making. Particularly, in virtual realitysystems for medical training, assessment of user´s skills isnecessary to know training quality and provide somefeedback about the user´s performance on criticalprocedures. An online assessment system for training insimulators based on virtual reality must have a lowcomplexity algorithm to do not compromise the performanceof the simulator. Several approaches to perform it have beenproposed. In this paper, it is proposed an approach foronline assessment based on a Naive Bayesian Network,where each variable is modeled by a Gaussian MixtureModel. This network was used for modeling andclassification of user’s skills into N a priori defined classesof performance.Index Terms ⎯  Training based on Virtual Reality, Users'Assessment, Medical Simulation, Gaussian Mixture Models.