Healthcare Analytics

HUI YANG, PhD, is Associate Professor in the Harold and Inge Marcus Department of Industrial and Manufacturing Engineering at The Pennsylvania State University. His research interests include sensor-based modeling and analysis of complex systems for process monitoring/control; system diagnostics/ prognostics; quality improvement; and performance optimization with special focus on nonlinear stochastic dynamics and the resulting chaotic, recurrence, self-organizing behaviors.EVA K. LEE, PhD, is Professor in the H. Milton Stewart School of Industrial and Systems Engineering at the Georgia Institute of Technology, Director of the Center for Operations Research in Medicine and HealthCare, and Distinguished Scholar in Health System, Health Systems Institute at both Emory University School of Medicine and Georgia Institute of Technology. Her research interests include health-risk prediction; early disease prediction and diagnosis; optimal treatment strategies and drug delivery; healthcare outcome analysis and treatment prediction; public health and medical preparedness; large-scale healthcare/medical decision analysis and quality improvement; clinical translational science; and business intelligence and organization transformation.

• List of Contributors xviiPreface xxiPart I Advances In Biomedical And Health Informatics 11 Recent Development in Methodology for Gene Network Problems and Inferences 3Sung W. Han and Hua Zhong1.1 Introduction 31.2 Background 51.3 Genetic Data Available 71.4 Methodology 71.5 Search Algorithm 131.6 PC Algorithm 151.7 Application/Case Studies 161.8 Discussion 231.9 Other Useful Softwares 232 Biomedical Analytics and Morphoproteomics: An Integrative Approach for Medical Decision Making for Recurrent or Refractory Cancers 31Mary F. McGuire and Robert E. Brown2.1 Introduction 312.2 Background 322.2.1 Data 332.2.2 Tools 332.2.3 Algorithms 342.2.4 Literature Review 352.3 Methodology 372.3.1 Morphoproteomics (Fig. 2.1(1–3)) 392.3.2 Biomedical Analytics (Fig. 2.1(4–10)) 402.3.3 Integrating Morphoproteomics and Biomedical Analytics 442.4 Case Studies 462.4.1 Clinical: Therapeutic Recommendations for Pancreatic Adenocarcinoma 462.4.2 Clinical: Biology Underlying Exceptional Responder in Refractory Hodgkin's Lymphoma 482.4.3 Research: Role of the Hypoxia Pathway in Both Oncogenesis and Embryogenesis 502.5 Discussion 512.6 Conclusions 523 Characterization and Monitoring of Nonlinear Dynamics and Chaos in Complex Physiological Systems 59Hui Yang, Yun Chen, and Fabio Leonelli3.1 Introduction 593.2 Background 613.3 Sensor-Based Characterization and Modeling of Nonlinear Dynamics 653.3.1 Multifractal Spectrum Analysis of Nonlinear Time Series 653.3.2 Recurrence Quantification Analysis 753.3.3 Multiscale Recurrence Quantification Analysis 783.4 Healthcare Applications 803.4.1 Nonlinear Characterization of Heart Rate Variability 813.4.2 Multiscale Recurrence Analysis of Space–Time Physiological Signals 853.5 Summary 884 Statistical Modeling of Electrocardiography Signal for Subject Monitoring and Diagnosis 95Lili Chen, Changyue Song, and Xi Zhang4.1 Introduction 954.2 Basic Elements of ECG 964.3 Statistical Modeling of ECG for Disease Diagnosis 994.3.1 ECG Signal Denoising 1004.3.2 Waveform Detection 1054.3.3 Feature Extraction 1064.3.4 Disease Classification and Diagnosis 1114.4 An Example: Detection of Obstructive Sleep Apnea from a Single ECG Lead 1154.4.1 Introduction to Obstructive Sleep Apnea 1154.5 Materials and Methods 1154.5.1 Database 1154.5.2 QRS Detection and RR Correction 1164.5.3 R Wave Amplitudes and EDR Signal 1174.5.4 Feature Set 1174.5.5 Classifier Training with Feature Selection 1184.6 Results 1184.6.1 QRS Detection and RR Correction 1184.6.2 Feature Selection 1184.6.3 OSA Detection 1204.7 Conclusions and Discussions 1215 Modeling and Simulation of Measurement Uncertainty in Clinical Laboratories 127Varun Ramamohan, James T. Abbott, and Yuehwern Yih5.1 Introduction 1275.2 Background and Literature Review 1295.2.1 Measurement Uncertainty: Background and Analytical Estimation 1305.2.2 Uncertainty in Clinical Laboratories 1345.2.3 Uncertainty in Clinical Laboratories: A System Approach 1365.3 Model Development Guidelines 1385.3.1 System Description and Process Phases 1385.3.2 Modeling Guidelines 1395.4 Implementation of Guidelines: Enzyme Assay Uncertainty Model 1415.4.1 Calibration Phase 1425.4.2 Sample Analysis Phase 1495.4.3 Results and Analysis 1505.5 Discussion and Conclusions 1526 Predictive Analytics: Classification in Medicine and Biology 159Eva K. Lee6.1 Introduction 1596.2 Background 1616.3 Machine Learning with Discrete Support Vector Machine Predictive Models 1636.3.1 Modeling of Reserved-Judgment Region for General Groups 1646.3.2 Discriminant Analysis via Mixed-Integer Programming 1656.3.3 Model Variations 1676.3.4 Theoretical Properties and Computational Strategies 1706.4 Applying DAMIP to Real-World Applications 1706.4.1 Validation of Model and Computational Effort 1716.4.2 Applications to Biological and Medical Problems 1716.5 Summary and Conclusion 1827 Predictive Modeling in Radiation Oncology 189Hao Zhang, Robert Meyer, Leyuan Shi, Wei Lu, and Warren D'Souza7.1 Introduction 1897.2 Tutorials of Predictive Modeling Techniques 1917.2.1 Feature Selection 1917.2.2 Support Vector Machine 1927.2.3 Logistic Regression 1937.2.4 Decision Tree 1937.3 Review of Recent Predictive Modeling Applications in Radiation Oncology 1947.3.1 Machine Learning for Medical Image Processing 1947.3.2 Machine Learning in Real-Time Tumor Localization 1967.3.3 Machine Learning for Predicting Radiotherapy Response 1977.4 Modeling Pathologic Response of Esophageal Cancer to Chemoradiotherapy 1997.4.1 Input Features 2007.4.2 Feature Selection and Predictive Model Construction 2007.4.3 Results 2027.4.4 Discussion 2047.5 Modeling Clinical Complications after Radiation Therapy 2057.5.1 Dose-Volume Thresholds: Relationship to OAR Complications 2057.5.2 Modeling the Radiation-Induced Complications via Treatment Plan Surface 2067.5.3 Modeling Results 2087.6 Modeling Tumor Motion with Respiratory Surrogates 2117.6.1 Cyberknife System Data 2117.6.2 Modeling for the Prediction of Tumor Positions 2127.6.3 Results of Tumor Positions Modeling 2127.6.4 Discussion 2147.7 Conclusion 2158 Mathematical Modeling of Innate Immunity Responses of Sepsis: Modeling and Computational Studies 221Chih-Hang J. Wu, Zhenzhen Shi, David Ben-Arieh, and Steven Q. Simpson8.1 Background 2218.2 System Dynamic Mathematical Model (SDMM) 2238.3 Pathogen Strain Selection 2248.3.1 Step 1: Kupffer Local Response Model 2248.3.2 Step 2: Neutrophils Immune Response Model 2288.3.3 Step 3: Damaged Tissue Model 2338.3.4 Step 4: Monocytes Immune Response Model 2348.3.5 Step 5: Anti-inflammatory Immune Response Model 2378.4 Mathematical Models of Innate Immunity of AIR 2398.4.1 Inhibition of Anti-inflammatory Cytokines 2398.4.2 Mathematical Model of Innate Immunity of AIR 2398.4.3 Stability Analysis 2418.5 Discussion 2478.5.1 Effects of Initial Pathogen Load on Sepsis Progression 2478.5.2 Effects of Pro- and Anti-inflammatory Cytokines on Sepsis Progression 2508.6 Conclusion 254Part II Analytics for Healthcare Delivery 2619 Systems Analytics: Modeling and Optimizing ClinicWorkflow and Patient Care 263Eva K. Lee, Hany Y. Atallah, Michael D. Wright, Calvin Thomas IV, Eleanor T. Post, Daniel T. Wu, and Leon L. Haley Jr9.1 Introduction 2649.2 Background 2669.3 Challenges and Objectives 2679.4 Methods and Design of Study 2689.4.1 ED Workflow and Services 2699.4.2 Data Collection and Time-Motion Studies 2709.4.3 Machine Learning for Predicting Patient Characteristics and Return Patterns 2749.4.4 The Computerized ED System Workflow Model 2779.4.5 Model Validation 2829.5 Computational Results, Implementation, and ED Performance Comparison 2859.5.1 Phase I: Results 2859.5.2 Phase I: Adoption and Implementation 2889.5.3 Phase II: Results 2889.5.4 Phase II: Adoption and Implementation 2909.6 Benefits and Impacts 2929.6.1 Quantitative Benefits 2949.6.2 Qualitative Benefits 2969.7 Scientific Advances 2979.7.1 Hospital Care Delivery Advances 2979.7.2 OR Advances 29810 A Multiobjective Simulation Optimization of the Macrolevel Patient Flow Distribution 303Yunzhe Qiu and Jie Song10.1 Introduction 30310.2 Literature Review 30510.2.1 Simulation Modeling on Patient Flow 30510.2.2 Multiobjective Patient Flow Optimization Problems 30610.2.3 Simulation Optimization 30710.3 Problem Description and Modeling 30810.3.1 Problem Description 30810.3.2 System Modeling 31010.4 Methodology 31210.4.1 Simulation Model Description 31210.4.2 Optimization 31310.5 Case Study: Adjusting Patient Flow for a Two-Level Healthcare System Centered on the Puth 31610.5.1 Background and Data 31610.5.2 Simulation under Current Situation 31810.5.3 Model Validation 32010.5.4 Optimization through Algorithm 1 32110.5.5 Optimization through Algorithm 2 32210.5.6 Comparison of the Two Algorithms 32710.5.7 Managerial Insights and Recommendations 32810.6 Conclusions and the Future Work 32911 Analysis of Resource Intensive Activity Volumes in US Hospitals 335Shivon Boodhoo and Sanchoy Das11.1 Introduction 33511.2 Structural Classification of Hospitals 33711.3 Productivity Analysis of Hospitals 33911.4 Resource and Activity Database for US Hospitals 34111.4.1 Medicare Data Sources for Hospital Operations 34311.5 Activity-Based Modeling of Hospital Operations 34411.5.1 Direct Care Activities 34411.5.2 The Hospital Unit of Care (HUC) Model 34711.5.3 HUC Component Results by State 35011.6 Resource use Profile of Hospitals from HUC Activity Data 35111.6.1 Comparing the Resource Use Profile of States 35311.6.2 Application of the Hospital Classification Rules 35511.7 Summary 35712 Discrete-Event Simulation for Primary Care Redesign: Review and a Case Study 361Xiang Zhong, Molly Williams, Jingshan Li, Sally A. Kraft, and Jeffrey S. Sleeth12.1 Introduction 36112.2 Review of Relevant Literature 36212.2.1 Literature on Primary Care Redesign 36212.2.2 Literature on Discrete-Event Simulation in Healthcare 36612.2.3 UW Health Improvement Projects 36912.3 A Simulation Case Study at a Pediatric Clinic 36912.3.1 Patient Flow 36912.3.2 Model Development 37112.3.3 Model Validation 37612.4 What–If Analyses 37612.4.1 Staffing Analysis 37612.4.2 Resident Doctor 37712.4.3 Schedule Template Change 37712.4.4 Volume Change 37912.4.5 Room Assignment 37912.4.6 Early Start 38012.4.7 Additional Observations 38212.5 Conclusions 38213 Temporal and Spatiotemporal Models for Ambulance Demand 389Zhengyi Zhou and David S. Matteson13.1 Introduction 38913.2 Temporal Ambulance Demand Estimation 39113.2.1 Notation 39213.2.2 Factor Modeling with Constraints and Smoothing 39313.2.3 Adaptive Forecasting with Time Series Models 39513.3 Spatiotemporal Ambulance Demand Estimation 39813.3.1 Spatiotemporal Finite Mixture Modeling 40013.3.2 Estimating Ambulance Demand 40313.3.3 Model Performance 40513.4 Conclusions 40914 Mathematical Optimization and Simulation Analyses for Optimal Liver Allocation Boundaries 413Naoru Koizumi, Monica Gentili, Rajesh Ganesan, Debasree DasGupta, Amit Patel, Chun-Hung Chen, Nigel Waters, and Keith Melancon14.1 Introduction 41414.2 Methods 41614.2.1 Mathematical Model: Optimal Locations of Transplant Centers and OPO Boundaries 41614.2.2 Discrete-Event Simulation: Evaluation of Optimal OPO Boundaries 42214.3 Results 42314.3.1 New Locations of Transplant Centers 42314.3.2 New OPO Boundaries 42614.3.3 Evaluation of New OPO Boundaries 42814.4 Conclusions 43315 Predictive Analytics in 30-Day Hospital Readmissions for Heart Failure Patients 439Si-Chi Chin, Rui Liu, and Senjuti B. Roy15.1 Introduction 44015.2 Analytics in Prediction Hospital Readmission Risk 44115.2.1 The Overall Prediction Pipeline 44115.2.2 Data Preprocessing 44115.2.3 Predictive Models 44215.2.4 Experiment and Evaluation 44415.3 Analytics in Recommending Intervention Strategies 44715.3.1 The Overall Intervention Pipeline 44715.3.2 Bayesian Network Construction 44815.3.3 Recommendation Rule Generation 45215.3.4 Intervention Recommendation 45315.3.5 Experiments 45415.4 Related Work 45715.5 Conclusion 45916 Heterogeneous Sensing and Predictive Modeling of Postoperative Outcomes 463Yun Chen, Fabio Leonelli, and Hui Yang16.1 Introduction 46316.2 Research Background 46616.2.1 Acute Physiology and Chronic Health Evaluation (APACHE) 46616.2.2 Simplified Acute Physiology Score (SAPS) 46916.2.3 Mortality Probability Model (MPM) 47016.2.4 Sequential Organ Failure Assessment (SOFA) 47216.3 Research Methodology 47416.3.1 Data Categorization 47516.3.2 Data Preprocessing and Missing Data Imputation 47516.3.3 Feature Extraction 48216.3.4 Feature Selection 48416.3.5 Predictive Model 48716.3.6 Cross-Validation and Ensemble Voting Processes 48916.4 Materials and Experimental Design 49116.5 Experimental Results 49116.6 Discussion and Conclusions 49817 Analyzing Patient–Physician Interaction in Consultation for Shared Decision Making 503Thembi Mdluli, Joyatee Sarker, Carolina Vivas-Valencia, Nan Kong, and Cleveland G. Shields17.1 Introduction 50317.2 Literature Review 50517.2.1 Patient–Physician Interaction on Prognosis Discussion 50617.2.2 Physician–Patient Interaction on Pain Assessment 50917.3 Our Recent Data Mining Studies 51017.3.1 Predicting Patient Satisfaction with Survey Data 51017.3.2 Predicting Patient Satisfaction with Conservation Data 51317.4 Future Directions 51517.4.1 Regression Shrinkage and Selection 51517.4.2 Conversational Characterization 51717.5 Concluding Remarks 51918 The History and Modern Applications of Insurance Claims Data in Healthcare Research 523Margrét V. Bjarndóttir, David Czerwinski, and Yihan Guan18.1 Introduction 52318.1.1 Advantages and Limitations of Claims Data 52518.1.2 Application Areas 52618.1.3 Statistical Methodologies Used in Claims-Based Studies 52818.2 Healthcare Cost Predictions 53118.2.1 Modeling of Healthcare Costs 53118.2.2 Modeling of Disease Burden and Interactions 53318.2.3 Performance Measures and Baselines 53418.2.4 Prediction Algorithms 53418.2.5 Applying Regression Trees to Cost Predictions 53518.2.6 Applying Clustering Algorithms to Cost Predictions 53718.2.7 Identifying High-Cost Members 53918.2.8 Discussion 53918.3 Measuring Quality of Care 54018.3.1 Structure, Process, and Outcomes 54018.3.2 The Quality of Quality Data 54218.3.3 Composite Quality Measures 54218.3.4 Practical Considerations for Constructing Quality Scores 54418.3.5 A Statistical Approach to Measuring Quality 54518.3.6 Quality as a Case Management Tool 54618.3.7 Discussion 54718.4 Conclusions 54819 Understanding the Role of Social Media in Healthcare via Analytics: a Health Plan Perspective 555Sinjini Mitra and Rema Padman19.1 Introduction 55519.2 Literature Review 55619.2.1 Privacy and Security Concerns in Social Media and Healthcare 55919.2.2 Analytics in Healthcare and Social Media 56119.3 Case Study Description 56219.3.1 Survey Design 56319.4 Research Methods and Analytics Tools 56419.4.1 The Logistic Regression Model 56419.5 Results and Discussions 56819.5.1 Descriptive Statistics 56819.5.2 Baseline of Technology Usage 57019.5.3 Mobile and Social Media Usage 57119.5.4 Clustering of Member Population by Technology, Social, and Mobile Media Usage 57219.5.5 Interest in Adopting Online Tools for Healthcare Purposes 57319.5.6 Interest in Adopting Mobile Apps for Healthcare Purposes 57419.5.7 Health and Wellness Objectives 57719.5.8 Privacy and Security Concerns 58019.5.9 Predictive Models 58119.6 Conclusions 584References 585Index 589