Machine Learning in Medicine – a Complete Overview

The current book is the first publication of a complete overview of machine learning methodologies for the medical and health sector. It was written as a training companion and as a must-read, not only for physicians and students, but also for any one involved in the process and progress of health and health care. In eighty chapters eighty different machine learning methodologies are reviewed, in combination with data examples for self-assessment. Each chapter can be studied without the need to consult other chapters.

The amount of data stored in the world’s databases doubles every 20 months, and clinicians, familiar with traditional statistical methods, are at a loss to analyze them. Traditional methods have, indeed, difficulty to identify outliers in large datasets, and to find patterns in big data and data with multiple exposure / utcome variables. In addition, analysis-rules for surveys and questionnaires, which are currently common methods of data collection, are, essentially, missing. Fortunately, the new discipline, machine learning, is able to cover all of these limitations.

So far medical professionals have been rather reluctant to use machine learning. Also, in the field of diagnosis making, few doctors may want a computer checking them, are interested in collaboration with a computer or with computer engineers. Adequate health and health care will, however, soon be impossible without proper data supervision from modern machine learning methodologies like cluster models, neural networks and other data mining methodologies.

Each chapter starts with purposes and scientific questions. Then, step-by-step analyses, using data examples, are given. Finally, a paragraph with conclusion, and references to the corresponding sites of three introductory textbooks, previously written by the same authors, is given.

Table of Contents

Part I: Cluster and Classification Models
Chapter 1: Hierarchical Clustering and K-Means Clustering to Identify Subgroups in Surveys (50 Patients)
Chapter 2: Density-Based Clustering to Identify Outlier Groups in Otherwise Homogeneous Data (50 Patients)
Chapter 3: Two Step Clustering to Identify Subgroups and Predict Subgroup Memberships in Individual Future Patients (120 Patients)
Chapter 4: Nearest Neighbors for Classifying New Medicines (2 New and 25 Old Opioids)
Chapter 5: Predicting High-Risk-Bin Memberships (1,445 Families)
Chapter 6: Predicting Outlier Memberships (2,000 Patients)
Chapter 7: Data Mining for Visualization of Health Processes (150 Patients)
Chapter 8: Trained Decision Trees for a More Meaningful Accuracy (150 Patients)
Chapter 9: Typology of Medical Data (51 Patients)
Chapter 10: Predictions from Nominal Clinical Data (450 Patients)
Chapter 11: Predictions from Ordinal Clinical Data (450 Patients)
Chapter 12: Assessing Relative Health Risks (3,000 Subjects)
Chapter 13: Measuring Agreement (30 Patients)
Chapter 14: Column Proportions for Testing Differences Between Outcome Scores (450 Patients)
Chapter 15: Pivoting Trays and Tables for Improved Analysis of Multidimensional Data (450 Patients)
Chapter 16: Online Analytical Procedure Cubes, a More Rapid Approach to Analyzing Frequencies (450 Patients)
Chapter 17: Restructure Data Wizard for Data Classified the Wrong Way (20 Patients)
Chapter 18: Control Charts for Quality Control of Medicines (164 Tablet Desintegration Times)

Part II: (Log) Linear Models
Chapter 19: Linear, Logistic, and Cox Regression for Outcome Prediction with Unpaired Data (20, 55, and 60 Patients)
Chapter 20: Generalized Linear Models for Outcome Prediction with Paired Data (100 Patients and 139 Physicians)
Chapter 21: Generalized Linear Models Event-Rates (50 Patients)
Chapter 22: Factor Analysis and Partial Least Squares (PLS) for Complex-Data Reduction (250 Patients)
Chapter 23: Optimal Scaling of High-Sensitivity Analysis of Health Predictors (250 Patients)
Chapter 24: Discriminant Analysis for Making a Diagnosis from Multiple Outcomes (45 Patients)
Chapter 25: Weighted Least Squares for Adjusting Efficacy Data with Inconsistent Spread (78 Patients)
Chapter 26: Partial Correlations for Removing Interaction Effects from Efficacy Data (64 Patients)
Chapter 27: Canonical Regression for Overall Statistics of Multivariate Data (250 Patients)
Chapter 28: Multinomial Regression for Outcome Categories (55 Patients)
Chapter 29: Various Methods for Analyzing Predictor Categories (60 and 30 Patients)
Chapter 30: Random Intercept Models for Both Outcome and Predictor Categories (55 patients)
Chapter 31: Automatic Regression for Maximizing Linear Relationships (55 patients)
Chapter 32: Simulation Models for Varying Predictors (9,000 Patients)
Chapter 33: Generalized Linear Mixed Models for Outcome Prediction from Mixed Data (20 Patients)
Chapter 34: Two-Stage Least Squares (35 Patients)
Chapter 35: Autoregressive Models for Longitudinal Data (120 Mean Monthly Population Records)
Chapter 36: Variance Components for Assessing the Magnitude of Random Effects (40 Patients)
Chapter 37: Ordinal Scaling for Clinical Scores with Inconsistent Intervals (900 Patients)
Chapter 38: Loglinear Models for Assessing Incident Rates with Varying Incident Risks (12 Populations)
Chapter 39: Loglinear Modeling for Outcome Categories (445 Patients)
Chapter 40: Heterogeneity in Clinical Research: Mechanisms Responsible (20 Studies)
Chapter 41: Performance Evaluation of Novel Diagnostic Tests (650 and 588 Patients)
Chapter 42: Quantile-Quantile Plots, a Good Start for Looking at Your Medical Data (50 Cholesterol Measurements and 58 Patients)
Chapter 43: Rate Analysis of Medical Data Better than Risk Analysis (52 Patients)
Chapter 44: Trend Tests Will Be Statistically Significant if Traditional Tests Are Not (30 and 106 Patients)
Chapter 45: Doubly Multivariate Analysis of Variance for Multiple Observations from Multiple Outcome Variables (16 Patients)
Chapter 46: Probit Models for Estimating Effective Pharmacological Treatment Dosages (14 Tests)
Chapter 47: Interval Censored Data Analysis for Assessing Mean Time to Cancer Relapse (51 Patients)
Chapter 48: Structural Equation Modeling (SEM) with SPSS Analysis of Moment Structures (Amos) for Cause Effect Relationships I  (35 Patients)
Chapter 49: Structural Equation Modeling (SEM) with SPSS Analysis of Moment Structures (Amos) for Cause Effect Relationships in Pharmacodynamic Studies II (35 Patients)

Part III: Rules Models
Chapter 50: Neural Networks for Assessing Relationships That Are Typically Nonlinear (90 Patients)
Chapter 51: Complex Samples Methodologies for Unbiased Sampling (9,678 Persons)
Chapter 52: Correspondence Analysis for Identifying the Best of Multiple Treatments in Multiple Groups (217 Patients)
Chapter 53: Decision Trees for Decision Analysis (1,004 and 953 Patients)
Chapter 54: Multidimensional Scaling for Visualizing Experienced Drug Efficacies (14 Pain-Killers and 42 Patients)
Chapter 55: Stochastic Processes for Long Term Predictions from Short Term Observations
Chapter 56: Optimal Binning for Finding High Risk Cut-­offs (1,445 Families)
Chapter 57: Conjoint Analysis for Determining the Most Appreciated Properties of Medicines to Be Developed (15 Physicians)
Chapter 58: Item Response Modeling for Analyzing Quality of Life with Better Precision (1,000 Patients)
Chapter 59: Survival Studies with Varying Risks of Dying (50 and 60 Patients)
Chapter 60: Fuzzy Logic for Improved Precision of Dose-­Response Data (8 Induction Dosages)
Chapter 61: Automatic Data Mining for the Best Treatment of a Disease (90 Patients)
Chapter 62: Pareto Charts for Identifying the Main Factors of Multifactorial Outcomes (2,000 Admissions to Hospital)
Chapter 63: Radial Basis Neural Networks for Multidimensional Gaussian Data (90 Persons)
Chapter 64: Automatic Modeling of Drug Efficacy Prediction (250 Patients)
Chapter 65: Automatic Modeling for Clinical Event Prediction (200 Patients)
Chapter 66: Automatic Newton Modeling in Clinical Pharmacology (15 Alfentanil Dosages, 15 Quinidine Time-Concentration Relationships)
Chapter 67: Spectral Plots for High Sensitivity Assessment of Periodicity (6 Years’ Monthly C Reactive Protein Levels)
Chapter 68: Runs Test for Identifying Best Regression Models (21 Estimates of Quantity and Quality of Patient Care)
Chapter 69: Evolutionary Operations for Process Improvement (8 Operation Room Air Condition Settings)
Chapter 70: Bayesian Networks for Cause Effect Modeling (600 Patients)
Chapter 71: Support Vector Machines for Imperfect Nonlinear Data (200 Patients with Sepsis)
Chapter 72: Multiple Response Sets for Visualizing Clinical Data Trends (811 Patient Visits)
Chapter 73: Protein and DNA Sequence Mining
Chapter 74: Iteration Methods for Crossvalidations (150 Patients with Pneumonia)
Chapter 75: Testing Parallel-Groups with Different Sample Sizes and Variances (5 Parallel-Group Studies)
Chapter 76: Association Rules Between Exposure and Outcome (50 and 60 Patients)
Chapter 77: Confidence Intervals for Proportions and Differences in Proportions (100 and 75 Patients)
Chapter 78: Ratio Statistics for Efficacy Analysis of New Drugs (50 Patients)
Chapter 79: Fifth Order Polynomes of Circadian Rhythms (1 Patient with Hypertension)
Chapter 80: Gamma Distribution for Estimating the Predictors of Medical Outcome Scores (110 Patients)