Bayesian biostatistics and diagnostic medicine

Lyle D. Broemeling1584887672, 9781584887676, 9781584887683

There are numerous advantages to using Bayesian methods in diagnostic medicine, which is why they are employed more and more today in clinical studies. Exploring Bayesian statistics at an introductory level, Bayesian Biostatistics and Diagnostic Medicine illustrates how to apply these methods to solve important problems in medicine and biology. After focusing on the wide range of areas where diagnostic medicine is used, the book introduces Bayesian statistics and the estimation of accuracy by sensitivity, specificity, and positive and negative predictive values for ordinal and continuous diagnostic measurements. The author then discusses patient covariate information and the statistical methods for estimating the agreement among observers. The book also explains the protocol review process for cancer clinical trials, how tumor responses are categorized, how to use WHO and RECIST criteria, and how Bayesian sequential methods are employed to monitor trials and estimate sample sizes. With many tables and figures, this book enables readers to conduct a Bayesian analysis for a large variety of interesting and practical biomedical problems.

---

Table of contents :
Bayesian Biostatistics and Diagnostic Medicine……Page 1
Contents……Page 3
Preface……Page 7
Acknowledgments……Page 9
Author……Page 10
1.2 Statistical Methods in Diagnostic Medicine……Page 11
1.3 Preview of Book……Page 12
1.5 Software……Page 14
References……Page 15
2.2 Imaging Modalities……Page 16
2.3 Activities in Diagnostic Imaging……Page 20
2.4 Accuracy and Agreement……Page 21
2.5 Developmental Trials for Imaging……Page 23
2.6 Protocol Review and Clinical Trials……Page 24
2.6.2 Phase I, II, and III Clinical Designs……Page 25
References……Page 27
3.2 Sentinel Lymph Node Biopsy for Melanoma……Page 29
3.3 Tumor Depth to Diagnose Metastatic Melanoma……Page 30
3.4 Biopsy for Nonsmall Cell Lung Cancer……Page 31
References……Page 32
4.1 Introduction……Page 34
4.2 Bayes Theorem……Page 35
4.3 Prior Information……Page 36
4.4 Posterior Information……Page 40
4.5.2 Estimation……Page 43
4.5.3.1 Introduction……Page 45
4.5.3.2 Binomial Example of Testing……Page 46
4.5.3.3 Comparing Two Binomial Populations……Page 47
4.6.1 Introduction……Page 48
4.6.2 A One- Sample Binomial for Response……Page 49
4.6.3 One- Sample Binomial with Prior Information……Page 51
4.7.1 Introduction……Page 52
4.7.2 Direct Methods of Computation……Page 53
4.7.3.1 Introduction……Page 56
4.7.3.2 Common Mean of Normal Populations……Page 57
4.7.3.3 MCMC Sampling with WinBUGS……Page 60
4.8 Exercises……Page 63
References……Page 64
5.1 Introduction……Page 66
5.2.2 Objectives……Page 67
5.2.4 Patient and Reader Selection……Page 68
5.2.5 Study Plan……Page 69
5.2.7 Statistical Design and Analysis……Page 70
5.2.8 References……Page 71
5.3.2 Classification Probabilities……Page 72
5.3.3 Predictive Values……Page 75
5.3.5 ROC Curve……Page 76
5.4.2 The Spokane Heart Study……Page 80
5.4.3 ROC Area……Page 81
5.4.4 Definition of the ROC Curve……Page 84
5.4.5 Choice of Optimal Threshold Value……Page 85
5.5.1 Introduction……Page 86
5.5.2 Bayesian ROC Curve for Clustered Information……Page 87
5.5.3 Clustered Data in Mammography……Page 89
5.6 Comparing Accuracy between Modalities……Page 91
5.7.1 Introduction……Page 94
5.7.2.1 Binary Tests……Page 95
5.7.2.2 Multinomial Outcomes……Page 98
5.7.3.1 One ROC Curve……Page 99
5.7.3.2 Two ROC Curves……Page 102
5.8 Exercises……Page 104
References……Page 106
6.1 Introduction……Page 107
6.2.2 Log Link Function……Page 108
6.2.3 Logistic Link……Page 109
6.2.4 Diagnostic Likelihood Ratio……Page 111
6.3.1 Introduction……Page 113
6.3.2 ROC Curve as Response to Therapy……Page 114
6.3.3 Diagnosing Prostate Cancer……Page 116
References……Page 117
7.1 Introduction……Page 119
7.2.1 Binary Scores……Page 120
7.2.2 Other Indices of Agreement……Page 122
7.2.4 Comparing Two Kappa Parameters……Page 123
7.2.5 Kappa and Stratification……Page 125
7.2.6 Multiple Categories and Two Readers……Page 126
7.2.7 Multiple Categories……Page 128
7.2.8 Agreement and Covariate Information……Page 129
7.3.1 Introduction……Page 132
7.3.2.1 One- Way Random Model……Page 133
7.3.2.2 Two- Way Random Model……Page 137
7.3.3 Regression and Agreement……Page 138
7.4 Combining Reader Information……Page 140
7.5 Exercises……Page 142
References……Page 145
8.1 Introduction……Page 146
8.2.2 Phase I Designs……Page 147
8.2.3 Phase II Trials……Page 148
8.3 Protocol……Page 150
8.4 Guidelines for Tumor Response……Page 151
8.5 Bayesian Sequential Stopping Rules……Page 153
8.6 Software for Clinical Trials……Page 157
8.6.2 Multc Lean for Phase II Trials……Page 158
8.7.1 Phase I Trial for Renal Cell Carcinoma……Page 159
8.7.2 An Ideal Phase II Trial……Page 161
8.7.3 Phase II Trial for Advanced Melanoma……Page 163
8.8 Exercises……Page 167
References……Page 168
9.1 Introduction……Page 169
9.2.1 Extreme Verification Bias……Page 170
9.2.2 Verification Bias……Page 174
9.2.3 Estimating Test Accuracy with No Gold Standard……Page 177
9.3 Test Accuracy and Survival Analysis……Page 182
9.4 ROC Curves with a Non- binary Gold Standard……Page 184
9.5.1 Inference for Sensitivity and Transition Probability……Page 186
9.5.2 Bayesian Inference for Lead- Time……Page 190
9.6 Decision Theory and Diagnostic Accuracy……Page 193
9.7 Exercises……Page 196
References……Page 197