Clustering of Order Sequences Based on the Typicalness Index for Finding Clinical Pathway Candidates

This paper presents a method for mining clinical pathway candidates from order history based on the typical ness index. Firstly, we constitute occurrence and transition frequency matrices of clinical orders based on the all cases. Next, we define the typical ness index of an order sequence based on the occurrence and transition frequencies and compute its value for each case. After that we perform clustering of all cases according to the similarity on the typical ness indices. Experimental results on an otorhinolaryngologic disease dataset demonstrate that the method is capable of producing clusters that reflect differences of treatment processes induced by the differences of operation dates.