Predictive data science is already in use in many fields, but its application in toxicology is new and sought after by non-animal alternative testing initiatives. Predictive Analytics for Toxicology: Applications in Discovery Science provides a comprehensive overview of the application of predictive analytics in the field of toxicology, highlighting its role and applications in discovery science.
This book addresses the challenges of accurately predicting high-level endpoints of toxicity and explores the use of computational and artificial intelligence research to automate predictive toxicology. It underscores the importance of predictive toxicology in proposing and explaining adverse outcomes resulting from human exposures to specific toxicants, especially when experimental and observational data on the toxicant are incomplete or unavailable.
Key features:
Includes a plain language description of predictive analytics in toxicology adding an overview of the wide range of applications
Examines the science of prediction, computational models as an automated science and comprehensive discussions on concepts of machine learning
Opens the hood on AI and its applications in toxicology
Features coverage on how in silico toxicity predictions are translational science tools
The book integrates strategies and practices of predictive toxicology and offers practical information that students and professionals of the toxicology, chemical, and pharmaceutical industries will find essential. It fulfills the expectations of student researchers seeking to learn predictive analytics in toxicology. This book will energize scientists to conduct predictive toxicology modeling using artificial intelligence and machine learning, and inspire students and seasoned scientists interested in automated science to pick up new research using predictive in silico models to evaluate chemical-induced toxicity.
With its focus on practical applications and real-world examples, this book serves as a guide for navigating the complex issues and practices of discovery toxicology. It is an essential resource for those interested in computer-based methods in toxicology providing valuable insights into the use of predictive analytics.
1. Finding No Limit: Discovery Science
1.1 Introduction
1.2 Predictive analytics
1.3 Data service
1.4 Examination of data quality
1.5 Program development intelligence
1.6 Data visualization
1.7 Decision analysis
1.8 Conclusions
2. Exploring the Science of Prediction
2.1 Introduction
2.2 The Ubiquitous Application of Prediction Science in Society
2.3 Understanding the Science of Prediction through Machine Learning
2.4 Predictive Modeling of the Human Brain
2.4.1 Human Learning and Logic
2.4.1.1 Inductive Reasoning
2.4.1.2 Deductive Reasoning
2.4.1.3 Abductive Reasoning
2.4.2 Computerized Modeling of Human Learning
2.5 Major Stages in Model Development and Prediction Testing
2.5.1 Model Objectives and Scope
2.5.2 Data Selection
2.5.3 Data Curation
2.5.4 Data Cleaning
2.5.5 Feature Selection
2.5.6 Model Construction
2.5.7 Model Testing
2.6 Current Trends in Machine Learning to Make Predictions
2.7 Conclusions
3. Predictive Analytic Approaches
3.1 Introduction
3.2 Descriptive Analytics
3.3 Diagnostic Analytics
3.4 Predictive Analytics
3.5 Prescriptive Analytics
3.6 Characteristics of a Predictive Model
3.7 Categories of Machine Learning Approaches
3.7.1 Supervised Machine Learning
3.7.2 Unsupervised Machine Learning
3.7.3 Semi-supervised Machine Learning
3.8 Machine Learning Algorithms
3.8.1 Unsupervised Machine Learning Techniques
3.8.1.1 Principal Component Analysis
3.8.1.2 Independent Component Analysis
3.8.1.3 Clustering
3.8.1.4 Self-organizing maps
3.8.2 Supervised Machine Learning
3.8.2.1 Linear Regression
3.8.2.2 Logistic Regression
3.8.2.3 Linear Discriminant Analysis
3.8.2.4 Decision Trees
3.8.2.5 Random Forest
3.8.2.6 Support Vector Machine
3.8.2.7 Naïve Bayes
3.8.2.8 k-Nearest Neighbours
3.8.2.9 Adaboost: Adaptive Boosting
3.8.2.10 Gradient Decent
3.8.2.11 Artificial Neural Networks and Deep Learning
3.8.2.11.1 Graph Neural Networks
3.9 Conclusions
4. The Role of Predictive Analytics in Translational Science
4.1 Introduction
4.2 What is Translational Science?
4.3 The Rapid Pace of Translational Science
4.4 Understanding Translational Science Research Tools
4.4.1 Relevant Targets for Translational Science Tools
4.5 Predictive Analytics in Advancing Translational Science
4.6 Development of Clinical Decision-Support Models
4.7 Conclusions
5. Toxicology and Artificial Intelligence
5.1 Introduction
5.2 Fundamental Ideas with AI
5.3 Toxicological Research with AI
5.3.1 Toxicogenomics
5.3.2 Disease and Health-based Research
5.3.3 Prediction of Toxicity
5.4 Limitations and Considerations with AI
5.4.1 AI and the Human Workforce
5.4.2 AI and Human Behavior
5.4.3 AI and Ethical Considerations
5.4.3.1 Bias
5.4.3.2 Safety and Trustworthiness
5.4.3.3 Model Fairness
5.4.3.4 Data Privacy
5.4.3.5 LLMs and Scientific Writing
5.4.3.6 Dual Use
5.4.4 An Environmental Dichotomy
5.5 Risk Assessment and AI
5.6 Conclusions
6. The Explosion of Computational Approaches to Hazard Identification
6.1 Introduction
6.2 Initiatives and Large-Scale Programs Supporting Computational Approaches for Hazard Identification of Chemicals
6.2.1 Toxicity Testing in the 21st Century: A Vision and Strategy - Tox21
6.2.1.1 Predictive Analytics Research in Response to Tox21
6.2.2 Other Initiatives and Major Activities Using Predictive Analytics Approaches for
Toxicology
6.3 Wide-Spread Availability of Public Data, Tools, and Resources for Modeling
6.4 Chemical Hazard Identification, Safety, and Risk-Driven Needs
6.5 Data Science Uses and Definition
6.6 Conclusions
7. Opportunities for Computational Modeling in Toxicology Prediction
7.1 Introduction
7.2 Current Application of Computational Modeling in Toxicity Prediction
7.3 Opportunistic Areas in Computational Modeling for Toxicology Prediction
7.3.1 Modeling High Value Organ-Specific Toxicities
7.3.1.1 Cardiac Toxicity Modeling
7.3.1.2 Drug-induced Liver Injury (DILI) Modeling
7.3.2 In Vitro Modeling
7.3.3 Mixtures Toxicity Modeling
7.3.4 Metabolic Reactions Modeling
7.3.5 Nanotoxicology Modeling
7.3.6 Modeling Based on Clinical Toxicology Data
7.3.7 Epidemiology Data-based Modeling
7.3.8 Advanced Artificial Intelligence (AI) Systems Modeling
7.4 Looking Forward
7.5 Conclusions
8. Key Challenges in Computational Modeling
8.1 Introduction
8.2 Data Availability and Quality
8.3 Model Overfitting and Underfitting
8.4 Feature Selection and Complex Dimensionality
8.4.1 Feature Selection Methods
8.4.2 Complex Dimensionality
8.4.3 Efficiency
8.5 Modeling Complex Chemical Interactions
8.6 Generalizability and Aging of Computational Models
8.7 Accuracy and Validation Assessment of Models
8.7.1 Validation and Verification
8.7.2 Uncertainty
8.8 Translating Models into Practice
8.9 Scalability
8.10 Expertise in Predictive Model Development and Use
8.10.1 Model selection
8.11 Education
8.12 Other Challenges
8.13 Conclusions
9. Predictive Data Science in Driving Growth in Toxicology
9.1 Introduction
9.2 Predictive Analytics in Toxicology
9.3 Advantages of Using Predictive Data Analytics in Toxicology
9.3.1 Cost and time efficiency
9.3.2 Enhanced accuracy
9.3.3 Reduction in use of animals for experimentation
9.3.4 Improved access to predictive data analytic tools
9.3.5 Data-driven approaches for better decision-making
9.4 Current Initiatives Fueling Predictive Data Analytics in Toxicology
9.5 Conclusions
Height:
Width:
Spine:
Weight:0.00