life science data mining

Transcription

1 life science data mining - '.)'-. < } ti» (>.:>,u» c ~'editors Stephen Wong Harvard Medical School, USA Chung-Sheng Li /BM Thomas J Watson Research Center World Scientific NEW JERSEY LONDON SINGAPORE. BEIJING SHANGHAI HONG KONG TAIPEI CHENNAI

2 CONTENTS Preface v Chapter 1 Survey of Early Warning Systems for Environmental and Public Health Applications 1 1. Introduction 1 2. Disease Surveillance 3 3. Reference Architecture for Model Extraction 5 4. Problem Domain 9 5. Data Sources Detection Methods Summary and Conclusion 13 References 14 Chapter 2 Time-Lapse Cell Cycle Quantitative Data Analysis Using Gaussian Mixture Models Introduction Material and Feature Extraction Material and cell feature extraction Model the time-lapse data using AR model Problem Statement and Formulation Classification Methods Gaussian mixture models and the EM algorithm K-Nearest Neighbor (KNN) classifier Neural networks Decision tree Fisher clustering i Experimental Results Trace identification Cell morphologic similarity analysis Phase identification Cluster analysis of time-lapse data 37 XI

3 xii Contents 6. Conclusion 40 Appendix A 41 Appendix B 42 References 43 Chapter 3 Diversity and Accuracy of Data Mining Ensemble Introduction Ensemble and Diversity Why needs diversity? Diversity measures Probability Analysis Coincident Failure Diversity Ensemble Accuracy Relationship between random guess and accuracy of lower bound single models Relationship between accuracy A and the number of models N When model's accuracy < 50% Construction of Effective Ensembles Strategies for increasing diversity Ensembles of neural networks Ensembles of decision trees Hybrid ensembles An Application: Osteoporosis Classification Problem Osteoporosis problem Results from the ensembles of neural nets Results from ensembles of the decision trees Results of hybrid ensembles Discussion and Conclusions 68 References 70 Chapter 4 Integrated Clustering for Microarray Data Introduction Related Work Data Preprocessing 81

4 Contents xiii 4. Integrated Clustering Clustering algorithms Integration methodology Experimental Evaluation Evaluation methodology Results Discussion Conclusions 94 References 94 Chapter 5 Complexity and Synchronization of EEG with Parametric Modeling Introduction Brief review of EEG recording analysis AR modeling based EEG analysis TVAR Modeling Complexity Measure Synchronization Measure Conclusions References 114 Chapter 6 Bayesian Fusion of Syndromic Surveillance with Sensor Data for Disease Outbreak Classification Introduction Approach Bayesian belief networks Syndromic data Environmental data Test scenarios Evaluation metrics Results Scenario Scenario Promptness Summary and Conclusions 136 References 137

5 xiv Contents Chapter 7 An Evaluation of Over-the-Counter Medication Sales for Syndromic Surveillance Introduction Background and Related Work Data : Approaches Lead-lag correlation analysis Regression test of predictive ability Detection-based approaches Supervised algorithm for outbreak detection in OTC data Modified Holt-Winters forecaster Forecasting based on multi-channel regression Experiments Lead-lag correlation analysis of OTC data Regression test of the predicative value of OTC Results from detection-based approaches Conclusions and Future Work 158 References 159 Chapter 8 Collaborative Health Sentinel Introduction Infectious Disease and Existing Health Surveillance Programs Elements of the Collaborative Health Sentinel (CHS) System Sampling Creating a national health map Detection Reaction Cost considerations Interaction with the Health Information Technology (HOT) World Conclusion 188 References 189 Appendix A - HL7 192

6 Contents xv Chapter 9 A Multi-Modal System Approach for Drug Abuse Research and Treatment Evaluation: Information Systems Needs and Challenges Introduction Context Data sources Examples of relevant questions Possible System Structure Challenges in System Development and Implementation Ontology development Data source control, proprietary issues Privacy, security issues Costs to implement/maintain system Historical hypothesis-testing paradigm Utility, usability, credibility of such a system Funding of system development Summary 207 References 208 Chapter 10 Knowledge Representation for Versatile Hybrid Intelligent Processing Applied in Predictive Toxicology Introduction Hybrid Intelligent Techniques for Predictive Toxicology Knowledge Representation XML Schemas for Knowledge Representation and Processing in AI and Predictive Toxicology Towards a Standard for Chemical Data Representation in Predictive Toxicology Hybrid Intelligent Systems for Knowledge Representation in Predictive Toxicology A formal description of implicit and explicit knowledge-based intelligent systems An XML schema for hybrid intelligent systems A Case Study Materials and methods Results 233

7 xvi Contents 7. Conclusions 235 References 236 Chapter 11 Ensemble Classification System Implementation for Biomedical Microarray Data Introduction Background Reasons for ensemble Diversity and ensemble Relationship between measures of diversity and combination method Measures of diversity Microarray data Ensemble Classification System (ECS) Design ECS overview Feature subset selection Base classifiers Combination strategy Experiments Experimental datasets Experimental results Conclusion and Further Work 254 References 255 Chapter 12 An Automated Method for Cell Phase Identification in High Throughput Time-Lapse Screens Introduction Nuclei Segmentation and Tracking Cell Phase Identification Feature calculation Identifying cell phase Correcting cell phase identification errors Experimental Results Conclusion 272 References 272

8 Contents xvii Chapter 13 Inference of Transcriptional Regulatory Networks Based on Cancer Microarray Data Introduction Subnetworks and Transcriptional Regulatory Networks Inference Inferring subnetworks using z-score Inferring subnetworks based on graph theory Inferring subnetworks based on Bayesian networks Inferring transcriptional regulatory networks based on integrated expression and sequence data Multinomial Probit Regression with Baysian Gene Selection Problem formulation Bayesian variable selection Bayesian estimation using the strongest genes Experimental results Network Construction Based on Clustering and Predictor Design Predictor construction using reversible jump MCMC annealing CoD for predictors Experimental results on amyeloid line Concluding Remarks 298 References 299 Chapter 14 Data Mining in Biomedicine Introduction Predictive Model Construction Derivation of unsupervised models Derivation of supervised models Validation Impact Analysis Summary 319 References 319

9 xviii Contents Chapter 15 Mining Multilevel Association Rules from Gene Ontology and Microarray Data Introduction Proposed Methods Preprocessing Hierarchy-information encoding The MAGO Algorithm MAGO algorithm CMAGO (Constrained Multilevel Association rules with Gene Ontology) Experimental Results The characteristic of thedataset Experimental results Interpretation Concluding Remarks 335 References 336 Chapter 16 A Proposed Sensor-Configuration and Sensitivity Analysis of Parameters with Applications to Biosensors Introduction Sensor-System Configuration Optical Biosensors Relationship between parameters Modelling of parameters Discussion Conclusion.?. 358 References 359 Epilogue 361 References.'. 364 Index 365