Nonhypothesis-Driven Research: Data Mining and Knowledge Discovery

摘要: Clinical information, stored over time, is a potentially rich source of data for clinical research. Knowledge discovery in databases (KDD), commonly known as mining, process pattern and predictive modeling large databases. KDD makes extensive use mining methods, automated processes, algorithms that enable recognition. Characteristically, involves the machine learning methods developed domain artificial intelligence. These have been applied to healthcare biomedical variety purposes with good success potential or realized translation. Herein, Fayyad model knowledge introduced. The steps are described select examples from research informatics. range initial selection interpretation evaluation. Commonly used surveyed: neural networks, decision tree induction, support vector machines (kernel methods), association rule k-nearest neighbor. Methods evaluating models result closely linked diagnostic medicine. include measures derived confusion matrix receiver operating characteristic curve analysis. Data partitioning validation critical aspects International efforts develop refine repositories critically these developing new knowledge.