Big Data Analysis and Artificial Intelligence for Medical Sciences

List of Contributors xiii Preface xix 1 Introduction 1 Bruno Carpentieri and Paola Lecca 1.1 Disease Diagnoses 4 1.2 Drug Development 6 1.3 Personalized Medicine 6 1.4 Gene Editing 7 Author Biographies 9 References 9 2 Fuzzy Logic for Knowledge-Driven and Data-Driven Modeling in Biomedical Sciences 17 Paolo Cazzaniga, Simone Spolaor, Caro Fuchs, Marco S. Nobile and Daniela Besozzi 2.1 Introduction 17 2.2 Fuzzy Logic 18 2.2.1 Fuzzy Sets 19 2.2.2 Linguistic Variables 19 2.2.3 Fuzzy Rules 20 2.2.4 Fuzzy Inference Systems 21 2.2.5 Simpful 22 2.3 Knowledge-Driven Modeling 22 2.3.1 Dynamic Fuzzy Modeling 23 2.3.2 Application 1: Maximizing Cancer Cells Death with Minimal Drug Combinations 25 2.3.3 FuzzX: A Hybrid Mechanistic-Fuzzy Modeling and Simulation Engine 27 2.3.4 Application 2: Analyzing Oscillatory Regimes in Signal Transduction Pathways 29 2.4 Data-Driven Modeling 30 2.4.1 pyFUME: Automatic Generation of Fuzzy Inference Systems 31 2.4.2 Application 3: Assessing Tremor Severity in Neurological Disorders 33 2.5 Discussion 35 Author Biographies 36 References 37 3 Application of Machine Learning Algorithms to Diagnosis and Prognosis of Chronic Wounds 43 Mai Dabas and Amit Gefen 3.1 Background 43 3.1.1 Chronic Wounds 43 3.1.2 Implementation of AI Methodologies in Wound Care and Management 43 3.2 Clinical Visual Assessment of Wounds Supported by Artificial Intelligence 44 3.2.1 Predicting the Formation and Progress of Wounds Based on Electronic Health Records 46 3.2.2 Predicting the Formation and Evolution of Wounds Based on a Dynamic Evaluation of Wound Characteristics and Relevant Physiological Measures 48 3.2.3 Feasible Implementation of AI Solutions For Wound Care Delivery and Management 49 3.2.4 Types of Data Modalities for Diagnosis, Detection, and Prediction of Chronic Wounds 50 3.3 Smartphone and Tablet Use in Wound Diagnosis and Management 51 3.4 Conclusions 53 Acronyms 54 Author Biographies 55 References 55 4 Deep Learning Techniques for Gene Identification in Cancer Prevention 59 Eleonora Lusito 4.1 The Next-Generation Era of Cancer Investigation 59 4.1.1 Cancer at Its First Definitions 59 4.1.2 Attempts to Sequence Nucleic Acids Over the Years 60 4.1.3 From the First to the Third-Generation Sequencing 61 4.1.4 Applications of NGS in Clinical Oncology 62 4.2 Deep Learning Approaches for Genomic Variants Identification in Cancer 63 4.2.1 Cancer Causing Factors 63 4.2.2 The Contribution of Germline Alterations to Cancer 64 4.2.3 Somatic Mutations and Cancer 64 4.2.4 Calling Variants from Sequence Data 65 4.2.5 Computational Approaches for Variant Discovery 65 4.2.6 Convolutional Neural Networks (CNNs): Basic Principles 66 4.2.7 Application of CNNs to Variant Calling 67 4.2.8 A Typical CNN Architecture for Variant Calling 68 4.2.9 The Activation Function 69 4.2.10 Dropout and L1–L2 Regularization 71 4.2.11 Advantages of Deep Learning Over the Existing Techniques 72 4.2.12 Residual Neural Networks (ResNet)-Inspired CNN in Genomic Variants Detection 73 4.3 Deep Learning in Cancer Transcriptomics 74 4.3.1 Gene Expression and Cancer 74 4.3.2 Analytical Approaches to Deal with Gene Expression Data 76 4.3.3 Stacked Denoising Autoencoders (SDAEs) for Dimensionality Reduction 76 4.3.4 The Variational Autoencoder (VAE) 79 4.3.5 VAEs to Integrate Gene Expression and Methylation Data 81 4.3.5.1 DNA Methylation: the Epigenetic Regulation of Gene Expression 81 4.3.5.2 Preprocessing Input Data of Different Sources 82 4.3.5.3 A VAE Architecture for Multimodal Data 82 4.4 Conclusions 84 Acronyms 86 Author Biographies 87 References 87 5 Deep Learning for Network Biology 97 Eleonora Lusito 5.1 Types of Interactions Between Genes and Their Products 97 5.2 Deep Learning Methods with Graph-input Data 99 5.2.1 Graph Embedding 99 5.2.1.1 Random Walk-Based Graph Embedding 100 5.2.1.2 Proximity-Based Graph Embedding 101 5.2.2 Graph Convolutional Networks (GCNs) 102 5.3 Applications of GNNs to Infer Biological and Pharmacological Interactions 104 5.3.1 Proteomics 104 5.3.2 Drug Development and Repurposing 104 5.3.3 Drug–Drug Interaction Prediction 105 5.3.4 Disease Classification and Outcome Prediction 106 Author Biography 107 References 107 6 Deep Learning-Based Reduced Order Models for Cardiac Electrophysiology 115 Stefania Fresca, Luca Dedè and Andrea Manzoni 6.1 Overview of Cardiac Physiology 115 6.1.1 Atrial Tachycardia and Atrial Fibrillation 117 6.1.2 Mathematical Models for Cardiac Electrophysiology 118 6.2 Reduced Order Modeling 121 6.2.1 Problem Formulation 123 6.2.2 Nonlinear Dimensionality Reduction 123 6.3 Decreasing Complexity in Cardiac Electrophysiology 124 6.3.1 POD-Enhanced Deep Learning-Based ROMs 125 6.3.1.1 POD-DL-ROM Architecture and Algorithms 128 6.4 Numerical Results 130 6.4.1 Test 1: Two-Dimensional Slab with Figure of Eight Reentry 131 6.4.2 Test 2: Three-Dimensional Left Ventricle Geometry 133 6.4.3 Test 3: Left Atrium Surface by Varying the Stimuli Location 135 6.4.4 Test 4: Reentry Breakup 137 6.5 Conclusions 139 Author Biographies 140 References 140 7 The Potential of Microbiome Big Data in Precision Medicine: Predicting Outcomes Through Machine Learning 149 Silvia Turroni and Simone Rampelli 7.1 The Gut Microbiome: A Major Player in Human Physiology and Pathophysiology 149 7.2 Machine Learning Applied to Microbiome Research 151 7.2.1 Case Study 1: Obesity 151 7.2.2 Case Study 2: Cancer 153 7.2.3 Case Study 3: Personalized Nutrition 154 7.2.4 Case Study 4: Exploiting the Meta-Community Theory for New Machine Learning Approaches 155 7.3 Conclusions and Perspectives 155 Author Biographies 156 References 156 8 Predictive Patient Stratification Using Artificial Intelligence and Machine Learning 161 Thanh-Phuong Nguyen, Thanh T. Giang, Quang T. Pham and Dang H. Tran 8.1 Overview of Artificial Intelligence for Patient Stratification 161 8.2 A RPCA and MKL Combination Model for Patient Stratification 164 8.2.1 Robust Principal Component Analysis 164 8.2.2 Dimensionality Reduction and Features Extraction Based on RPCA 166 8.2.3 Predictive Model Construction Based on Multiple Kernel Learning 168 8.2.4 Materials 169 8.2.4.1 Cancer Patient Datasets 169 8.2.4.2 Alzheimer Disease Patient Datasets 170 8.2.5 Experiment Design 171 8.2.5.1 Experiment of Stratifying Cancer Patients 171 8.2.5.2 Experiment of Stratifying Alzheimer Disease Patients 171 8.2.6 Results and Discussions 171 8.2.6.1 Application of Stratifying Cancer Patients 172 8.2.7 Application of Stratifying Alzheimer Disease Patients 174 8.3 Conclusion 175 Author Biographies 175 References 176 9 Hybrid Data-Driven and Numerical Modeling of Articular Cartilage 181 Seyed Shayan Sajjadinia, Bruno Carpentieri and Gerhard A. Holzapfel 9.1 Introduction 181 9.2 Knee and Cartilage 182 9.2.1 Main Joint Substructures 182 9.2.2 Load-Bearing Cartilage Phases 183 9.3 Physics-Based Modeling 185 9.3.1 Numerical Modeling 185 9.3.2 Constitutive Modeling 188 9.4 AI-Enhanced Modeling 191 9.4.1 Deep Learning 191 9.4.2 Surrogate Modeling 192 9.5 Discussion and Conclusion 194 Author Biographies 194 References 195 10 A Hybrid of Differential Evolution and Minimization of Metabolic Adjustment for Succinic and Ethanol Production 205 Zhang N. Hor, Mohd S. Mohamad, Yee W. Choon, Muhammad A. Remli and Hairudin A. Majid 10.1 Introduction 205 10.2 Method 206 10.2.1 Differential Evolution (DE) 206 10.2.2 Mutation 206 10.2.3 Crossover 207 10.2.4 Selection 208 10.2.5 Minimization of Metabolic Adjustment 208 10.2.6 A Hybrid of Differential Evolution and Minimization of Metabolic Adjustment 209 10.3 Experiments and Discussion 209 10.3.1 Dataset 209 10.3.2 Parameter Setting 209 10.3.3 Experimental Results 210 10.3.4 Comparative Analysis 214 10.4 Conclusion 214 Acknowledgment 215 Author Bibliographies 215 References 216 11 Analysis Pipelines and a Platform Solution for Next-Generation Sequencing Data 219 Víctor Duarte, Alesandro Gómez and Juan M. Corchado 11.1 Introduction 219 11.2 NGS Data Analysis Pipeline and State of the Art Tools 220 11.2.1 Quality Assessment 220 11.2.2 Alignment 221 11.2.3 Post-alignment and pre-variant Calling Processing 222 11.2.4 Variant Calling 223 11.2.5 Variant Annotation 228 11.3 Nanopore Sequencing Data Analysis 229 11.3.1 Base-Calling 230 11.3.2 Quality Control and Preprocessing 230 11.3.3 Error Correction 231 11.3.4 Alignment 231 11.3.5 Variant Calling 231 11.4 Machine Learning Approaches in Variant Calling 232 11.5 Next-Generation Sequencing Data Analysis Frameworks 233 11.6 DeepNGS 235 11.6.1 Pipeline 235 11.6.2 DeepNGS Main Features 236 11.6.2.1 Power and Speed 236 11.6.2.2 Optimized Workflow 236 11.6.2.3 Intuitive Design and Interactive Charts 237 11.6.2.4 Extended Information 237 11.6.2.5 Artificial Intelligence and Machine Learning 237 11.7 Conclusions 240 Author Biographies 241 References 241 12 Artificial Intelligence: From Drug Discovery to Clinical Pharmacology 253 Paola Lecca 12.1 Artificial Intelligence and the Druggable Genome 253 12.2 Feature-Based Methods 257 12.3 Similarity/Distance-Based Methods 257 12.4 Matrix Factorization 258 12.4.1 Causal K-Nearest-Neighborhood 261 12.4.2 Causal Random Forests 263 12.4.3 Causal Support Vector Machine 264 12.5 Opportunities and Challenges 265 Author Biography 266 References 266 13 Using AI to Steer Brain Regeneration: The Enhanced Regenerative Medicine Paradigm 273 Gabriella Panuccio, Narayan P. Subramaniyam, Angel Canal-Alonso, Juan M. Corchado and Carlo Ierna 13.1 The Challenge of Brain Regeneration 273 13.2 The Enhanced Regenerative Medicine Paradigm 274 13.3 The Case of Epilepsy 276 13.4 AI to Understand Epilepsy 279 13.4.1 Commonly Applied Learning Algorithms for Basic Neuroscience and Clinical Application in Epilepsy 282 13.4.2 Seizure and Epilepsy Type Classification 284 13.4.3 Seizure Onset Zone Localization 284 13.4.4 Seizure Detection 285 13.4.5 Seizure Prediction 285 13.4.6 Signal Feature Extraction for Seizure Detection and Prediction 288 13.4.7 Network Interactions and Evolving Dynamics in the Epileptic Brain: The Eye of AI 290 13.5 Artificial Intelligence to Guide Graft-Host Dynamics in Epilepsy 292 13.6 Challenges and Limitations 294 13.6.1 From AI to Explainable AI 295 13.7 A Philosophical Perspective on Enhanced Brain Regeneration 297 Acknowledgments 299 Acronyms 299 Author Biographies 300 References 300 14 Towards Better Ways to Assess Predictive Computing in Medicine: On Reliability, Robustness, and Utility 309 Federico Cabitza and Andrea Campagner 14.1 Introduction 309 14.2 On Ground Truth Reliability 311 14.2.1 Weighted Reliability 314 14.2.2 Example Application 316 14.3 On Utility Metrics to Evaluate ML Performance 318 14.3.1 Weighted Utility 318 14.3.2 Example Application 321 14.4 On the Replicability of Clinical ML Models 322 14.4.1 Dataset Size 323 14.4.2 Dataset Similarity 325 14.4.3 Meta-Validation Procedure 325 14.4.4 Example Application 328 14.5 Conclusions and Future Outlook 331 Author Biographies 332 References 333 15 Legal Aspects of AI in the Biomedical Field. The Role of Interpretable Models 339 Chiara Gallese 15.1 Introduction 339 15.2 Data Protection 340 15.3 Transparency Principle 343 15.3.1 Right of Explanation 343 15.3.2 Right of Information 348 15.3.3 Informed Consent Requirements 349 15.4 Accountability Principle 350 15.5 Non-discrimination Principle and Biases 351 15.6 High-Risk Systems and Human Oversight 353 15.7 Additional Requirements of the AI Act Proposal 354 15.8 Interpretability as a Standard 355 15.9 Conclusion 358 Author Biography 358 References 359 16 The Long Path to Usable AI 363 Barbara Di Camillo, Enrico Longato, Erica Tavazzi and Martina Vettoretti 16.1 Promises and Challenges of Artificial Intelligence in Healthcare 363 16.2 Deployment of Usable Artificial Intelligence Models 367 16.2.1 Case Study: Predicting the Cardiovascular Complications of Diabetes via a Deep Learning Approach 368 16.3 Potential and Challenges of Employing Longitudinal Clinical Data in AI 375 16.3.1 Case Study: Modeling the Progression of Amyotrophic Lateral Sclerosis Through a Dynamic Bayesian Network 378 16.3.2 Case Study: Investigating Amyotrophic Lateral Sclerosis Progression Trajectories Leveraging Process Mining 381 16.4 Enhancing the Applicability of AI Predictive Models by a Combined Model Approach: A Case Study on T2D Onset Prediction 386 16.4.1 The Problem of Type 2 Diabetes Prediction 386 16.4.2 Potential Applications of T2D Predictive Models 387 16.4.3 Barriers to the Adoption of T2D Predictive Models 387 16.4.4 Addressing Practical Issues by Combining Multiple T2D Predictive Models 388 16.4.5 The Combined Model Achieves High Prediction Performance with High Coverage 390 16.5 Conclusions and Future Outlook 391 Author Biography 392 References 393 Index 399