Modern electronic health records are designed to capture and render vast quantities of clinical data during the health care process. Technological advancements in the form of computer-based patient records software and personal computer hardware are making the collection of and access to health care data more manageable. However, few tools exist to evaluate and analyze this clinical data after it has been captured and stored. Evaluation of stored clinical data may lead to discovery of trends and patterns hidden within the data that could significantly enhance our understanding of disease progression and management. A common goal of the medical data mining is the detection of some kind of correlation, for example, between genetic features and phenotypes or between medical treatment and reaction of patients (Abidi & Goh, 1998; Li et al., 2005). The characteristics of clinical data, including issues of data availability and complex representation models, can make data mining applications challenging.
Knowledge discovery in databases (KDD) is defined as the nontrivial extraction of implicit, previously unknown, and potentially useful information from data (Adriaans & Zantinge, 1996; Han & Kamber, 2001). Data mining is one step in the KDD where a discovery-driven data analysis technique is used for identifying patterns and relationships in datasets. Recent advances in medical science have led to revolutionary changes in medical research and technology and the accumulation of a large volume of medical data that demands in-depth analysis. The question becomes how to bridge the two fields, data mining and medical science, for an efficient and successful mining of medical data.
While data analysis and data mining methods have been extensively applied for industrial and business applications, their utilization in medicine and health care is sparse (Abadi & Goh, 1998; Babic, 1999; Brossette, Sprague, Hardin, Jones, & Moser, 1998). In Ohsaki, Yoshinori, Shinya, Hideto, and Takahira (2003), the authors discuss the methods of obtaining medically valuable rules and knowledge on pre- and post-processing and the interaction between system and human expert using the data of medical tests results on chronic hepatitis. They developed the system based on the combination of pattern extraction with clustering and classification with decision tree and generated graph-based rules to predict prognosis. In Tsumoto (2000), the author focuses on the characteristics of medical data and discusses how data miner deals with medical data. In (Ohsaki et. al., 2007), authors discuss the usefulness of the interestingness measures for medical data mining through experiments using clinical datasets on meningitis. Based on the outcomes of these experiments, they discuss how to utilize these measures in postprocessing.
The data mining techniques such as Neural Network, Naïve Bayes, and Association rules are at present not well explored on medical databases. We are in the process of experimenting with a data mining project using gastritis data from Howard University Hospital in Washington, DC to identify factors that contribute to this disease. This project implements a wide spectrum of data mining techniques. The eventual goal of this data mining effort is to identify factors that will improve the quality and cost effectiveness of patient care.
In this article, we discuss the challenges facing the medical data mining. We present and analyze our experimental results on gastritis database by employing different data mining techniques such as Neural Network, Naive Bayes, and Association rules and using the data mining tool XLMiner (Shmueli, Patel, & Bruce, 2007; XLMiner, 2007).
Key Terms in this Chapter
Data Mining: Data mining is one step in the knowledge Discovery in Databases (KDD) where a discovery-driven data analysis technique, such as Naïve Bayes or Neural Networks or Association rules, is used for identifying patterns and relationships in data sets
XLMiner: XLMiner is a comprehensive data mining add-in for Excel. It offers a variety of methods to analyze data. It has extensive coverage of statistical and machine learning techniques for classification, prediction, affinity analysis and data exploration and reduction
Knowledge Discovery: Knowledge discovery in databases (KDD) is defined as the nontrivial extraction of implicit, previously unknown, and potentially useful information from data
Neural Networks Classification: Artificial neural networks process records one at a time, and “learn” by comparing their prediction of the record (which, at the outset, is largely arbitrary) with the known actual record. The errors from the initial prediction of the first record is fed back into the network, and used to modify the networks algorithm the second time around, and so on for many iterations
Classif ication: Discriminant Analysis, Naïve Bayes, Neural Network, Classification Tree, Logistic Regression, K-Nearest Neighbor
Prediction: Multiple Linear Regressions, K-Nearest Neighbor, Regression Tree, Neural Networks
Naïve Bayes Classification: Bayesian classifiers operate by saying “If you see a fruit that is red and round, which type of fruit is it most likely to be, based on the training data set? In future, classify red and round fruit as that type of fruit.”
Association Rules: Association rule mining finds interesting associations or correlation relationships among large set of data items. Association rules show attributes value conditions that occur frequently together in a given dataset. Association rules provide information of this type in the form of “if-then” statements.
Data Mining Techniques: The general categories of the data mining techniques include:
Data Preparation: The data preparation stage involved two stages: data cleaning and transformation. The data cleaning step eliminated inconsistent data. The transformation aims at converting the data fields to numeric fields or vice versa.
Affinity: Association rules