In this paper we describe results of an experimental study where we checked the impact of various difficulty factors in imbalanced data sets on the performance of selected classifiers applied alone or combined with several preprocessing methods. In the study we used artificial data sets in order to systematically check factors such as dimensionality, class imbalance ratio or distribution of specific types of examples (safe, borderline, rare and outliers) in the minority class. The results revealed that the latter factor was the most critical one and it exacerbated other factors (in particular class imbalance). The best classification performance was demonstrated by non-symbolic classifiers, particular by k-NN classifiers (with 1 or 3 neighbors - 1NN and 3NN, respectively) and by SVM. Moreover, they benefited from different preprocessing methods - SVM and 1NN worked best with undersampling, while oversampling was more beneficial for 3NN.
A prominent characteristic of clinical data is their heterogeneity—such data include structured examination records and laboratory results, unstructured clinical notes, raw and tagged images, and genomic data. This heterogeneity poses a formidable challenge while constructing diagnostic and therapeutic decision models that are currently based on single modalities and are not able to use data in different formats and structures. This limitation may be addressed using data fusion methods. In this paper, we describe a case study where we aimed at developing data fusion models that resulted in various therapeutic decision models for predicting the type of treatment (surgical vs. non-surgical) for patients with bone fractures. We considered six different approaches to integrate clinical data: one fusion model based on combination of data (COD) and five models based on combination of interpretation (COI). Experimental results showed that the decision model constructed following COI fusion models is more accurate than decision models employing COD. Moreover, statistical analysis using the one-way ANOVA test revealed that there were two groups of constructed decision models, each containing the set of three different models. The results highlighted that the behavior of models within a group can be similar, although it may vary between different groups.