Artificial Intelligence for Drug Development, Precision Medicine, and Healthcare

Cover
Half Title
Title Page
Copyright Page
Table of Contents
Preface
1: Overview of Modern Artificial Intelligence
	1.1 Brief History of Artificial Intelligence
	1.2 Waves of Artificial Intelligence
		1.2.1 First Wave: Logic-Based Handcrafted Knowledge
		1.2.2 Second Wave: Statistical Machine Learning
		1.2.3 Third Wave: Contextual Adaptation
		1.2.4 The Last Wave: Artificial General Intelligence
	1.3 Machine Learning Methods
		1.3.1 Data Science
		1.3.2 Supervised Learning: Classification and Regression
		1.3.3 Unsupervised Learning: Clustering and Association
		1.3.4 Reinforcement Learning
		1.3.5 Swarm Intelligence
		1.3.6 Evolutionary Learning
	1.4 Summary
	1.5 Problems
2: Classical Statistics and Modern Machine Learning
	2.1 Essentials for Medical Data Scientists
		2.1.1 Structured and Unstructured Data
		2.1.2 Random Variation and Its Causes
		2.1.3 Internal and External Validities
		2.1.4 Placebo and Nocebo Effects
		2.1.5 Bias, Bias, and Bias
		2.1.6 Confounding Factors
		2.1.7 Regression to the Mean
	2.2 Revolutionary Ideas of Modern Clinical Trials
		2.2.1 Innovative and Adaptive Development Program
		2.2.2 Control, Blinding, and Randomization
	2.3 Hypothesis Test and Modeling in Classic Statistics
		2.3.1 Statistical Hypothesis Testing
		2.3.2 Generalized Linear Model
		2.3.3 Air Quality Analysis with Generalized Linear Model
		2.3.4 Lung Cancer Survival Analysis with Cox's Model
		2.3.5 Propensity Score Matching
	2.4 Model Selection in Machine Learning
		2.4.1 Decision Approach
		2.4.2 Regularization
		2.4.3 Subset Selection
		2.4.4 Real-World Examples
	2.5 Process to Apply Machine Learning to Data
		2.5.1 General Steps in Applying Machine Learning
		2.5.2 Cross-Validation
	2.6 Summary
	2.7 Problems
3: Similarity Principle|The Fundamental Principle of All Sciences
	3.1 Scientific Paradoxes Call for a New Approach
		3.1.1 Dilemma of Totality Evidence with p-Value
		3.1.2 Multiple-Testing Versus Multiple-Learning
		3.1.3 A Medical and Judicial Tragedy
		3.1.4 Simpson's Paradox
		3.1.5 Bias in Predicting Drug Effectiveness
	3.2 The Similarity Principle
		3.2.1 Role of Similarity Principle
		3.2.2 The Root of Causality
	3.3 Similarity Measures
		3.3.1 Attributes Selection
		3.3.2 Similarity Properties
		3.3.3 Cosine Similarity and Jaccard Index
		3.3.4 Distance-Based Similarity Function
		3.3.5 Similarity and Dissimilarity of String and Signal Data
		3.3.6 Similarity and Dissimilarity for Images and Colors
		3.3.7 Similarix
		3.3.8 Adjacency Matrix of Network
		3.3.9 Biological and Medical Similarices
	3.4 Summary
	3.5 Problems
4: Similarity-Based Artificial Intelligence
	4.1 Similarity-Based Machine Learning
		4.1.1 Nearest-Neighbors Method for Supervised Learning
		4.1.2 Similarity-Based Learning
		4.1.3 Similarity Measures
		4.1.4 Algorithms for SBML
		4.1.5 Prediction Error Decomposition
		4.1.6 Training, Validation, and Test Datasets
	4.2 Regularization and Cross-Validation
		4.2.1 Learning|Updating Attribute-Scaling Factors
		4.2.2 Loss Function
		4.2.3 Computer Implementation
	4.3 Case Studies
	4.4 Different Outcome Variables
	4.5 Further Development of Similarity-Based AI Approach
		4.5.1 Repeated Measures
		4.5.2 Missing Data Handling
		4.5.3 Multiple Outcomes
		4.5.4 Sequential Similarity-Based Learning
		4.5.5 Ensemble Methods and Collective Intelligence
		4.5.6 Generalized SBML
		4.5.7 Dimension Reduction
		4.5.8 Recursive SBML
	4.6 Similarity Principle, Filtering, and Convolution
	4.7 Summary
	4.8 Problems
5: Artificial Neural Networks
	5.1 Hebb's Rule and McCulloch-Pitts Neuronal Model
	5.2 The Perceptron
		5.2.1 Model Construction
		5.2.2 Perceptron Learning
		5.2.3 Linear Separability
	5.3 Multiple-Layer Perceptron for Deep Learning
		5.3.1 Model Construction
		5.3.2 Gradient Method
	5.4 Artificial Neural Network with R
		5.4.1 ANN for Infertility Modeling
		5.4.2 Feedforward Network with Karasr Package
		5.4.3 MNIST Handwritten Digits Recognition
	5.5 Summary
	5.6 Problems
6: Deep Learning Neural Networks
	6.1 Deep Learning and Software Packages
	6.2 Convolutional Neural Network for Deep Learning
		6.2.1 Ideas Behind CNN
		6.2.2 Network Scalability Problem
		6.2.3 Deep Learning Architecture
		6.2.4 Illustration of CNN with Example
		6.2.5 CNN for Medical Image Analysis
		6.2.6 A CNN for Handwritten Digits Recognition
		6.2.7 Training CNN Using Keras in R
	6.3 Recurrent Neural Networks
		6.3.1 Short-Term Memory Network
		6.3.2 An Example of RNN in R
		6.3.3 Long Short-Term Memory Networks
		6.3.4 Sentiment Analysis Using LSTMs in R
		6.3.5 Applications of LSTMs in Molecular Design
	6.4 Deep Belief Networks
		6.4.1 Restricted Boltzmann machine
		6.4.2 Application of Deep Belief Networks
	6.5 Generative Adversarial Networks
	6.6 Autoencoders
	6.7 Summary
	6.8 Problems
7: Kernel Methods
	7.1 Subject Representation Using Kernels
	7.2 Prediction as Weighted Kernels
	7.3 Support Vector Machine
		7.3.1 Hard-Margin Model
		7.3.2 Soft-Margin Model
		7.3.3 R Program for Support Vector Machine
	7.4 Feature and Kernel Selections
	7.5 Application of Kernel Methods
	7.6 Dual Representations
	7.7 Summary
	7.8 Problems
8: Decision Tree and Ensemble Methods
	8.1 Classification Tree
	8.2 Regression Tree
	8.3 Bagging and Boosting
	8.4 Random Forests
	8.5 Summary
	8.6 Problems
9: Bayesian Learning Approach
	9.1 Bayesian Paradigms
	9.2 Bayesian Networks
		9.2.1 Bayesian Network for Molecular Similarity Search
		9.2.2 Coronary Heart Disease with Bayesian Network
	9.3 Bayesian Inference
		9.3.1 Basic Formulations
		9.3.2 Preclinical Study of Fluoxetine on Time Immobile
	9.4 Model Selection
	9.5 Hierarchical Model
	9.6 Bayesian Decision-Making
	9.7 Summary and Discussion
	9.8 Problems
10: Unsupervised Learning
	10.1 Needs of Unsupervised Learning
	10.2 Association or Link Analysis
	10.3 Principal Components Analysis
	10.4 K-Means Clustering
	10.5 Hierarchical Clustering
	10.6 Self-Organizing Maps
	10.7 Network Clustering and Modularity
	10.8 Unsupervised to Supervised Learning
	10.9 Summary
	10.10 Problems
11: Reinforcement Learning
	11.1 Introduction
	11.2 Sequential Decision-Making
		11.2.1 Descriptive and Normative Decision-Making
		11.2.2 Markov Chain
		11.2.3 Markov Decision Process
		11.2.4 Dynamic Programming
	11.3 Pharmaceutial Decision Process
		11.3.1 Model for Clinical Development Program
		11.3.2 Markov Decision Tree and Out-Licensing
	11.4 Q-Learning
	11.5 Bayesian Stochastic Decision Process
	11.6 Partially Observable Markov Decision Processes
	11.7 Summary
	11.8 Problems
12: Swarm and Evolutionary Intelligence
	12.1 Swarm Intelligence|Artificial Ants
		12.1.1 Artificial Swarm Intelligence
		12.1.2 Applications
	12.2 Evolutionary Intelligence
		12.2.1 Genetic Algorithm
		12.2.2 Genetic Algorithm for Infertility
		12.2.3 Genetic Programming
		12.2.4 Application
	12.3 Cellular Automata
	12.4 Summary
	12.5 Problems
13: Applications of AI in Medical Science and Drug Development
	13.1 AI for QSARs in Drug Discovery
		13.1.1 Deep Learning Networks
		13.1.2 Network Similarity-Based Machine Learning
		13.1.3 Kernel Method and SVMs
		13.1.4 Decision-Tree Method
		13.1.5 Other AI Methods
		13.1.6 Comparisons with Different Methods
	13.2 AI in Cancer Prediction Using Microarray Data
		13.2.1 Cancer Detection from Gene Expression Data
		13.2.2 Feature Selection
		13.2.3 Cancer Prediction
		13.2.4 Clustering
	13.3 Deep Learning for Medical Image Analysis
		13.3.1 Deep Learning for Medical Image Processing
		13.3.2 Deep Learning Methods in Mammography
		13.3.3 Deep Learning for Cardiological Image Analysis
	13.4 AI in Healthcare
		13.4.1 Paradigm Shift
		13.4.2 Disease Diagnosis and Prognosis
		13.4.3 Natural Language Processing in Medical Records
	13.5 AI for Clinical Trial and Drug Safety Monitoring
		13.5.1 Necessary Paradigm Shift in Clinical Trials
		13.5.2 Learning Paradigms
		13.5.3 AI in Pharmacovigilance
	13.6 Summary
14: Future Perspectives—Artificial General Intelligence
15: Appendix
	15.1 Data for Learning Artificial Intelligence
	15.2 AI Software Packages
	15.3 Derivatives of Similarity Functions
	15.4 Derivation of Backpropagation Algorithms for ANN
	15.5 Similarity-Based Machine Learning in R
Bibliography
Index