Automatic sleep scoring using integrated rule- and case-based reasoning

We suggest an automated method for interpreting polysomno signals by rule- and case-based reasoning system. The main idea of the research is to use the rules to generate an approximate assessment for the target subject, and to use cases to handle exceptions to the rules caused by the sleep complexity. The system first performs RBR based on the R and K standard, and then supplements the rules with CBR. Such integration combines both deductive and inductive approaches to automatic sleep diagnosis. We applied this approach to several cases to test its feasibility. In the normal case it worked better than using single RBR.