Artificial Intelligence for Healthy Longevity

Description

This book reviews the state-of-the-art efforts to apply machine learning and AI methods for healthy aging and longevity research, diagnosis, and therapy development. The book examines the methods of machine learning and their application in the analysis of big medical data, medical images, the creation of algorithms for assessing biological age, and effectiveness of geroprotective medications. The promises and challenges of using AI to help achieve healthy longevity for the population are manifold. This volume, written by  world-leading experts working at the intersection of AI and aging, provides a unique synergy of these two highly prominent fields and aims to create a balanced and comprehensive overview of the application methodology that can help achieve healthy longevity for the population. The book is accessible and valuable for specialists in AI and longevity research, as well as a wide readership, including gerontologists, geriatricians, medical specialists, and students from diverse fields, basic scientists, public and private research entities, and policy makers interested in potential intervention in degenerative aging processes using advanced computational tools.         Preface Contents Contributors Part I Biomarkers of Aging and Health 1 AI in Longevity 1.1 Statistical Models of Aging 1.1.1 Deep Learning Basic Principles 1.2 Deep Learning Aging Clocks 1.3 Deep Learning Applications in Medicine 1.3.1 Clinical Practice 1.3.2 Drug Development References 2 Automated Reporting of Medical Diagnostic Imaging for Early Disease and Aging Biomarkers Detection 2.1 Longevity Medicine and Radiology 2.2 AI and Radiology: State of the Art 2.3 Proof of Concept 2.4 Future of AI in Longevity Medicine References 3 Risk Forecasting Tools Based on the Collected Information for Two Types of Occupational Diseases 3.1 Development of a Mathematical Model Based on Risk Calculation Methodology Using Artificial Intelligence Tools 3.2 Overview of the Methods Used 3.3 Feature Description of Objects 3.4 Separation of Features by the Type of Their Values 3.5 Separation of Features by Meaning 3.6 The Type of the Problem to Be Solved and the Target Variables 3.7 Trained Models and Their Results 3.8 Linear Regression Model 3.9 Determining the Risk of Sensorineural Hearing Loss 3.10 Determination of the Risk of Vibration Disease (Local Vibration) 3.11 Determination of the Risk of Vibration Disease (General Vibration) 3.12 The Decision Tree Model 3.13 Determining the Risk of Sensorineural Hearing Loss 3.14 Determination of the Risk of Vibration Disease (Local Vibration) 3.15 Determination of the Risk of Vibration Disease (General Vibration) 3.16 Random Forest Model 3.17 Determining the Risk of Sensorineural Hearing Loss 3.18 Determination of the Risk of Vibration Disease (Local Vibration) 3.19 Determination of the Risk of Vibration Disease (General Vibration) 3.20 Predictive Risk Model 3.21 Conclusions and Model Selection 3.22 An Example of an Explicitly Interpreted DSS Fragment Based on an Automated Generated and Optimized Model 3.23 Justification of Risk Prediction Error Estimation 3.24 The First Approach to Estimating the Risk Prediction Error 3.25 The Second Approach to Estimating the Error of the Risk Forecast 3.26 Conclusions References 4 Obtaining Longevity Footprints in DNA Methylation Data Using Different Machine Learning Approaches 4.1 Introduction 4.2 Biological Age Regression with Machine Learning Models 4.2.1 Baseline Models 4.2.2 One-Tissue Epigenetic Clocks 4.2.3 Pan-Tissue Epigenetic Clocks 4.3 Machine Learning for Age-Related Diseases Classification 4.3.1 Cancer Classification 4.3.2 Phenotype Classification 4.3.3 Case–Control Classification 4.4 Unsupervised Learning for Cancer Differentiating 4.5 Conclusions References 5 The Role of Assistive Technology in Regulating the Behavioural and Psychological Symptoms of Dementia 5.1 Introduction 5.2 Methodology 5.3 Results 5.3.1 Assistive Technologies and Dementia 5.3.2 Assistive Technology to Aid Communication 5.3.3 Assistive Technology to Aid Motor Behaviour 5.3.4 Assistive Technology to Aid Inappropriate Behaviours 5.3.5 Assistive Technology—Smart Homes 5.3.6 Assistive Technology—Further Artificial Intelligence 5.4 Discussion 5.5 Conclusion References 6 Epidemiology, Genetics and Epigenetics of Biological Aging: One or More Aging Systems? 6.1 Introduction 6.2 An Overview of Aging Clocks Based on Biometric and Molecular Data 6.2.1 Telomere Length 6.2.2 DNA Methylation Aging Clocks 6.2.3 Blood Biomarker-Based Aging Clocks 6.2.4 Neuroimaging Based Brain Aging Clocks 6.3 Overlap of Aging Clocks 6.3.1 Epidemiological Overlap 6.3.2 Biological Overlap 6.4 Conclusions References 7 Temporal Relation Prediction from Electronic Health Records Using Graph Neural Networks and Transformers Embeddings 7.1 Introduction 7.2 Methods 7.2.1 Graph Construction 7.2.2 Masked Language Modeling 7.2.3 Model 7.3 Results 7.4 Discussion and Future Work References 8 In Silico Screening of Life-Extending Drugs Using Machine Learning and Omics Data 8.1 Introduction 8.2 Methods 8.2.1 Data Collection 8.2.2 Lines of Invertebrate Models and Keeping Conditions 8.2.3 Statistics and Reproducibility 8.2.4 Model 8.3 Results 8.3.1 Model Validation on Synthetic Data 8.3.2 Lifespan Tests 8.4 Discussion References 9 An Overview of Kernel Methods for Identifying Genetic Association with Health-Related Traits 9.1 Introduction 9.2 Introduction to the Kernels Methods for Genomic Data Analysis 9.2.1 The Kernel Functions 9.2.2 The Main Idea of the Kernel Methods 9.2.3 The Kernel Trick 9.2.4 Distance Induced by the Function Kernel 9.3 Kernel Machine Regression for Multi-marker genetic Association Testing 9.3.1 Kernel Linear and Logistic Regression Models for Genetic Association Test 9.3.2 Kernel Linear Regression Models 9.3.3 Kernel Logistic Regression Model 9.3.4 Rare-Variants Association Genetic Test 9.3.5 Connection with Other Multi-markers Association Tests 9.4 Selection of Variables for Gene-Set Analysis Using Kernels Methods 9.5 Kernel Methods for Association Genetics Test with Multiples Phenotypes 9.5.1 Genetic Association Test for Multiple Phenotypes Analysis Based on Kernel Methods 9.5.2 Multivariate Kernel Machine Regression (MKMR) 9.5.3 Multi-trait Sequence Kernel Association Test (MSKAT) 9.5.4 Gene Association with Multiple Traits (GAMuT) 9.6 Kernel Methods for Censored Survival Outcomes in Genetics Association Studies 9.7 Conclusions References 10 Artificial Intelligence Approaches for Skin Anti-aging and Skin Resilience Research 10.1 Introduction 10.2 Genetics 10.2.1 Skin Aging Genes 10.2.2 Long-Lived Individuals and Twin Studies 10.2.3 Telomere Shortening 10.2.4 Variety of Aging Clocks 10.3 Molecular Aging 10.3.1 Omics Approaches in the Aging Research 10.4 Drug Discovery for Skin Resilience 10.5 Microbiome 10.6 Skin 3D Models 10.7 Biophysical Markers 10.8 Skin Imaging 10.8.1 Skin Image and Video Processing 10.8.2 Morphology Features and aging Patterns 10.8.3 AI Systems for Facial Imaging 10.8.4 Discoloration 10.8.5 Skin Texture 10.8.6 Nails and Hair 10.8.7 Estimation of Age 10.8.8 Simulation of Aging 10.9 Psychology 10.10 The Future of Skin Anti-aging Is Personalized References Part II Perspectives and Challenges in Machine Learning Research of Aging and Longevity 11 AI in Genomics and Epigenomics 11.1 AI to Diagnose Monogenic Diseases 11.1.1 AI Helps to Call Genomic Variants from Massive Parallel Sequencing Data 11.1.2 AI for Clinical Interpretation of Genomic Variants 11.1.3 Interpretation of Genomic Data and Clinical Description of Patient’s Phenotype 11.2 Interpretation of Epigenetic Changes in Aging References 12 The Utility of Information Theory Based Methods in the Research of Aging and Longevity 12.1 Introduction 12.2 Definitions and Applications of Information-Theoretical Methods for the Research of Aging and Aging-Related Ill Health 12.3 The Application of Information-Theoretical Methods for the Evaluation of Biological and Biomedical Boundaries or Thresholds 12.4 Using Information-Theoretical Methods for Risk Group Attribution 12.5 Utilizing Information-Theoretical Methods for Genomic Sequence Analysis 12.6 Conclusion References 13 AI for Longevity: Getting Past the Mechanical Turk Model Will Take Good Data References 14 Leveraging Algorithmic and Human Networks to Cure Human Aging: Holistic Understanding of Longevity via Generative Cooperative Networks, Hybrid Bayesian/Neural/Logical AI and Tokenomics-Mediated Crowdsourcing 14.1 Introduction 14.2 Aging as a Complex Network Process 14.3 The Generative Cooperative Network (GCN) 14.4 Emergent Signs in GCN 14.5 How Emergent Signs Interact with Dependent Typing in the GCN 14.6 Data Absorption in GCN 14.7 BayesExpert in GCN 14.8 How BayesExpert can Combine Separate Studies into Coherent Wholes 14.9 Model Combination via Quadratic Programming 14.10 Hand-Crafted Bayes Net Model of Individual Aging 14.11 OpenCog’s BioAtomspace 14.12 Types of Atoms in the BioAtomspace 14.13 Tokenomic Incentivization 14.14 Strategies for Addressing Longevity via the Crowdsourced GCN Meta-Model References Author Index Subject Index