P-Course: medical applications of Bayesian classification with informative priors
Jussi Lahtinen, Petri Myllymäki, Olli‐Pekka Ryynänen · 2006
Abstract. B-Course is a web-based data analysis service for Bayesian modeling. Although already in use by researchers from many scientific disciplines, the current version of B-Course does not allow domain experts to incorporate prior knowledge into the system, but the models are constructed from sample data only by using machine learning techniques. P-Course is an attempt to modify B-Course so that domain knowledge can be expressed in form of priors, which are then combined with the statistical data. The current version of P-Course is designed for supervised classification problems. We discuss the design principles of the P-Course system and demonstrate its usefulness by using real-world examples from the medical domain. 1