Data mining of mental health issues of non-bone marrow donor siblings

Table 1 Conceptual differences of data mining approach.

Research area	Electronic medical record	Genomics/Proteomics	This study: Mental health on NDS
Data source	Physicians/nurses' Description, laboratory data and radiologic images on medical record	Gene expression data from cDNA microarray/mass spectrometry	Interview with the subject
Expected results	Automatic and effective data extraction/sorting	Extraction of genes/proteins with statistical significance Classification of gene/proteins Visualization of gene/protein expression pattern or pathway	Extraction of important and rarely-appeared words Visualization of relationship between keywords
Concept*	Supervised/Unsupervised approach	Supervised/Unsupervised approach	Unsupervised approach
Representative algorism of data mining technique	Data extraction matching with prepared data criteria	To provide statistically meaningful analysis for high-throughput and multi-dimensional biological data in the association with phenotype	To discover unanticipated, rarely appeared key-elements by Scenario Map analysis
Aims	Linking between medical record description and research issues To develop effective and commonly available electronic health record	To discover new biomarker or diagnostic method To discover therapeutic target	For better clinical follow-up by understanding unanticipated individual concerns

Conceptual differences of data mining approach in representative medical research areas are shown. *Supervised approach aims for testing or validation of hypothesis while unsupervised approach used for discovering unanticipated events or knowledge.