[2I1-GS-2] Machine learning: Random forest

[2I1-GS-2-04] Machine Learning based Prediction of DPC from Discharge Summaries

〇Tomohiro Kimura¹, Shusaku Tsumoto², Shoji Hirano² (1. Division of Medical Service, Faculty of Medicine, Shimane University, 2. Department of Medical Informatics, Faculty of Medicine, Shimane University)

Keywords:Text mining, Random Forest, Deep Learning, Discharge Summary

This paper proposes a method for construction of classifiers for discharge summaries, composed of the following five steps First, morphological analysis is applied to a set of summaries and a term matrix is generated. Second, correspondence analysis is applied to the classification labels and the term matrix and generates two dimensional coordinates for all the terms and labels. Third, by measuring the distances between categories and the terms, ranking of key words is generated. Fourthly, keywords are selected as attributes according to the ranks, and training examples for classifiers will be generated. Finally, machine learning methods are applied to the training examples. Experimental validation shows that random forest achieved the best performance and the second best was the deep learners, but decision tree methods with many keywords performed only a little worse than neural network or deep learning methods.