Search

Japanese medical device adverse events terminology, published by the Japan Federation of Medical Devices Associations (JFMDA terminology), contains entries for 89 terminology items, with each of the terminology entries created independently. It is necessary to establish and verify the consistency of these terminology entries and map them efficiently and accurately. Therefore, developing an automatic synonym detection tool is an important concern. Such tools for edit distances and distributed representations have achieved good performance in previous studies. The purpose of this study was to identify synonyms in JFMDA terminology and evaluate the accuracy using these algorithms. A total of 125 definition sentence pairs were created from the terminology as baselines. Edit distances (Levenshtein and Jaro–Winkler distance) and distributed representations (Word2vec, fastText, and Doc2vec) were employed for calculating similarities. Receiver operating characteristic analysis was carried out to evaluate the accuracy of synonym detection. A comparison of the accuracies of the algorithms showed that the Jaro–Winkler distance had the highest sensitivity, Doc2vec with DM had the highest specificity, and the Levenshtein distance had the highest value in area under the curve. Edit distances and Doc2vec makes it possible to obtain high accuracy in predicting synonyms in JFMDA terminology.

In this paper, we present an evaluation of learning costs of rule evaluation models based on objective indices for an iterative rule evaluation support method in data mining post-processing. Post-processing of mined results is one of the key processes in a data mining process. However, it is difficult for human experts to find out valuable knowledge from several thousands of rules obtained with a large dataset with noises. To reduce the costs in such rule evaluation task, we have developed the rule evaluation support method with rule evaluation models, which learn from objective indices for mined classification rules and evaluations by a human expert for each rule. To estimate learning costs for predicting human interests with objective rule evaluation indices, we have done the two case studies with actual data mining results, which include different phases of human interests. With regarding to these results, we discuss about the relationship between performances of learning algorithms and human hypothesis construction process.

In this paper, we present an evaluation of learning costs of rule evaluation models based on objective indices for an iterative rule evaluation support method in data mining postprocessing. Post-processing of mined results is one of the key processes in a data mining process. However, it is difficult for human experts to completely evaluate several thousands of rules from a large dataset with noises. To reduce the costs in such rule evaluation task, we have developed the rule evaluation support method with rule evaluation models, which learn from objective indices for mined classification rules and evaluations by a human expert for each rule. To estimate learning costs for predicting human interests with objective rule evaluation indices, we have done the two case studies with actual data mining results, which include different phases of human interests. With regard to these results, we discuss about learning costs to predict real human interests with objective rule evaluation indices.

A machine learning technique called Graph-Based Induction (GBI) efficiently extracts typical patterns from graph-structured data by stepwise pair expansion (pairwise chunking). It is very efficient because of its greedy search. We have expanded GBI to construct a decision tree that can handle graph-structured data. DT-GBI constructs a decision tree while simultaneously constructing attributes for classification using GBI. In DT-GBI attributes, namely substructures useful for classification task, are constructed by GBI on the fly during the tree construction. We applied both GBI and DT-GBI to classification tasks of a real world hepatitis data. Three classification problems were solved in five experiments. In the first 4 experiments, DT-GBI was applied to build decision trees to classify 1) cirrhosis and non-cirrhosis (Experiments 1 and 2), 2) type C and type B (Experiment 3), and 3) positive and negative responses of interferon therapy (Experiment 4). As the patterns extracted in these experiments are thought discriminative, in the last experiment (Experiment 5) GBI was applied to extract descriptive patterns for interferon therapy. The preliminary results of experiments, both constructed decision trees and their predictive accuracies as well as extracted patterns, are reported in this paper. Some of the patterns match domain experts' experience and the overall results are encouraging.

We have proposed a method called Decision Tree Graph-Based Induction (DT-GBI), which constructs a classifier (decision tree) for graph-structured data while simultaneously constructing attributes for classification. Graph-Based Induction (GBI) is utilized in DT-GBI for efficiently extracting typical patterns from graph-structured data by stepwise pair expansion (pairwise chunking). Attributes, i.e., substructures useful for classification task, are constructed by GBI on the fly while constructing a decision tree in DT-GBI. We applied DT-GBI to four classification tasks of hepatitis data using only the time-series data of blood inspection and urinalysis, which was provided by Chiba University Hospital. In the first and second experiments, the stages of fibrosis were used as classes and a decision tree was constructed for discriminating patients with F4 (cirrhosis) from patients with the other stages. In the third experiment, the types of hepatitis (B and C) were used as classes, and in the fourth experiment the effectiveness of interferon therapy was used as class label. The preliminary results of experiments, both constructed decision trees and their predictive accuracies, are reported in this paper. The validity of extracted patterns is now being evaluated by the domain experts (medical doctors). Some of the patterns match experts' experience and the overall results are encouraging.

We propose a method which visualizes irregular multidimensional time-series data as a sequence of probabilistic prototypes for detecting exceptions from medical test data. Conventional visualization methods often require iterative analysis and considerable skill thus are not totally supported by a wide range of medical experts. Our PrototypeLines displays summarized information based on a probabilistic mixture model by using hue only thus is considered to exhibit novelty. The effectiveness of the summarization is pursued mainly through use of a novel information criterion. We report our endeavor with chronic hepatitis data, especially discoveries of interesting exceptions by a nonexpert and an untrained expert.

This research empirically investigates the performance of conventional rule interestingness measures and discusses their practicality for supporting KDD through human-system interaction in medical domain. We compared the evaluation results by a medical expert and those by selected measures for the rules discovered from a dataset on hepatitis. Recall, Jaccard, Kappa, CST, X2-M, and Peculiarity demonstrated the highest performance, and many measures showed a complementary trend under our experimental conditions. These results indicate that some measures can predict really interesting rules at a certain level and that their combinational use will be useful.