Artificial Intelligence in Covid-19

Foreword
Preface
Contents
About the Editors
Chapter 1: Introduction to Artificial Intelligence in COVID-19
Pandemics
History of Pandemics
The COVID-19 Pandemic
Origins of the COVID-19 Pandemic
Continuous Fight for Science and Reason
Modern Tools for Pandemic Control
A Brief Chronology of the Chapters of This Book
Power of Science
References
Chapter 2: AI for Pooled Testing of COVID-19 Samples
Introduction
System Model
The PCR Process
Mathematical Model
Pooled COVID-19 Tests
Recovery from Pooled Tests
Group Testing Methods for COVID-19
Adaptive GT Methods
Non-Adaptive GT Methods
Pooling Matrix
Noiseless Linear Non-Adaptive Recovery
Noisy Non-Linear Non-Adaptive Recovery
Summary
Compressed Sensing for Pooled Testing for COVID-19
Compressed Sensing Forward Model for Pooled RT-PCR
CS Algorithms for Recovery
Details of Algorithms
Assessment of Algorithm Performance and Experimental Protocols
Choice of Pooling Matrices
Choice of Number of Pools
Use of Side Information in Pooled Inference
Comparative Discussion and Summary
References
Chapter 3: AI for Drug Repurposing in the Pandemic Response
Introduction
Desirable Features of AI for Drug Repurposing in Pandemic Response
Technical Flexibility and Efficiency
Clinical Applicability and Acceptability
Major AI Applications for Drug Repurposing in Response to COVID-19
Knowledge Mining
Network-Based Analysis
In Silico Modelling
IDentif.AI Platform for Rapid Identification of Drug Combinations
Project IDentif.AI
IDentif.AI for Drug Optimization Against SARS-CoV-2
IDentif.AI 2.0 Platform in an Evolving Pandemic
IDentif.AI as a Pandemic Preparedness Platform
Use of Real-World Data to Identify Potential Targets for Drug Repurposing
Future Directions
References
Chapter 4: AI and Point of Care Image Analysis for COVID-19
Introduction
Motivation for Using Imaging
Motivation for Using AI with Imaging
Integration of Imaging with Other Modalities
Literature Overview
Chest X-Ray Imaging
Diagnosis Models
Prognosis Models
Use of Longitudinal Imaging
Fusion with Other Data Modalities
Common Issues with AI and Chest X-Ray Imaging
Duplication and Quality Issues
Source Issues
Frankenstein Datasets
Implicit Biases in the Source Data
Artificial Limitations Due to Transfer Learning
Computed Tomography Imaging
Diagnosis Models
Prognosis Models
Applications to Regions Away from the Lungs
Use of Longitudinal Imaging
Fusion with Other Data Modalities
Common Issues with AI and Computed Tomography Imaging
Ultrasound Imaging
What Can be Observed in LUS
Models Assisting in Interpreting LUS
Diagnosis Models
Prognosis Models
Use of Longitudinal Imaging
Common Issues with AI and Ultrasound Imaging
Conclusions
Success Stories
Pitfalls to Focus On
Lessons Learned and Recommendations
The Next Pandemic
References
Chapter 5: Machine Learning and Laboratory Values in the Diagnosis, Prognosis and Vaccination Strategy of COVID-19
Introduction
COVID-19, Machine Learning and Laboratory Values: The State of the Art
Literature Search Results
Diagnostic Studies
Prognostic Studies
Considerations on the Literature Reviewed
Heterogeneity in Patient Selection
Laboratory Parameters Used by Machine Learning Models
Types of Models and Their Validation
Model Implementation
The Role of Artificial Intelligence in the Vaccination Strategy Against SARS-COV-2 Through Laboratory Tests
Real-World Vaccination Strategies
Artificial Intelligence Potentialities
Conclusions
Appendix 1
Diagnostic Papers (D)
Prognostic Papers (P)
Appendix 2: Tool Online
References
Chapter 6: AI and the Infectious Medicine of COVID-19
Introduction
AI and ML for SARS-CoV-2 Early Research Using Pathogen Sequence Data
AI and ML for Research of SARS-CoV-2 Antivirals
AI and ML for COVID-19 Infectious Medicine Early Research Using Language Data
AI and ML in Real World Data Analysis of COVID-19
AI and ML in Molecular Diagnostics of COVID-19
AI and ML in Image-Based Diagnostics of COVID-19 and Clinical Decision Support
AI and ML in COVID-19 Medical Care
Prevention, Infection Risk and Epidemiology
Treatment and Prognosis
Conclusions
References
Chapter 7: AI and ICU Monitoring on the Wake of the COVID-19 Pandemic
Introduction
ICU Monitoring Through AI
ICU Monitoring and AI in Pre-pandemic Times
The Impact of the COVID-19 Pandemic on the ICU and the Role of AI
Conclusions
References
Chapter 8: Symptom Based Models of COVID-19 Infection Using AI
Introduction
Using Machine Learning Methods to Determine Mortality of Patient with COVID-19
Using Machine Learning Methods to Detect the Presence of COVID-19 Infection
Using Machine Learning Methods to Differentiate COVID-19 and Influenza/Common Cold Infections
Summary, Limitations, Challenges, and Future Applications
References
Chapter 9: AI Techniques for Forecasting Epidemic Dynamics: Theory and Practice
Introduction
A Review of Model Types and Limits to Forecasting
Preliminaries
Model Details
Metrics for Forecast Evaluations
AI-Driven Engineering
An Example of a Real-time Forecasting Model
Results
A GNN-Based Spatio-Temporal Model
Additional Details Regarding the Framework
Forecasting Performance
Theoretical Foundations for Forecasting in Network Models
Overview
Some Short-Term Forecasting Problems and Their Computational Intractability
Discussion
References
Chapter 10: Regulatory Aspects on AI and Pharmacovigilance for COVID-19
What Does Artificial Intelligence Mean According to Legal Definition?
AI and Health
The European Union Legal Framework: A Work in Progress
The Proposed EU Regulation (Artificial Intelligence Act)
The Use of AI in Research and Developing Medicinal Products and Monitoring Their Quality, Safety and Efficacy
The Added Value Brought Using Artificial Intelligence in Performing Pharmacovigilance Activities in General and During the COVID-19 Pandemic
Ethical Issues: A Few Caveats
The Personal Data Protection Implications
Provisional Conclusions
Suggested Reading
Chapter 11: AI and the Clinical Immunology/Immunoinformatics for COVID-19
Introduction
Challenge for Traditional Vaccines in COVID-19
Long Development and Design Period
Difficulties in Knowing and Optimizing the Efficacy and Side Effects
Uncertainties with the Development and Other Costs During Production, Storage, and Transportation
Hard to Tackle Unknown and Emerging Mutations of Viruses
Existing AI Techniques Help the Traditional Vaccine Development in COVID-19
AI Makes the Practical Experimental Results Computational
AI-Based Computational Tools Can Help the Traditional Vaccine Design
AI-Based In Silico Vaccine Design
Our Recently Proposed DeepVacPred Vaccine Design Framework
Artificial Intelligence for Investigating Viral Evolution and Mutations
An Algorithmic Information Theoretic Approach to Discover the State Machine Generator Governing the Viral Sequence Structure and Enabling AI Strategies for Viral Mutation Prediction
Characterizing the Temporal Evolution of SARS-CoV-2 in a Continuous Manner
Detecting Regions Within Viral Sequences Likely to Exhibit Mutations
Summary
References
Chapter 12: AI and Dynamic Prediction of Deterioration in Covid-19
Introduction
COVID-19: A Novel Disease—Usage of Newer or Older Clinical Decisions Support Systems?
Clinical Decisions Support System Stable Parameters/Features Using Threshold Values
Patient Deterioration
General Prediction Scores
Early Warning Systems (EWS)
AI for Prediction of Deterioration
AI Assisted Patient-Specific Risk Prediction
AI Assisted Prediction of Critical Illness and Deterioration in COVID-19 Patients
Mortality Prediction Models for Covid-19
Mortality Prediction Models Using High-Frequency Data
Prediction Models for Sepsis
Explainable and Interpretable Machine Learning Methods for Clinical Decision Support Systems
References
Chapter 13: AI, Epidemiology and Public Health in the Covid Pandemic
Introduction
Epidemiology: Definition and Purposes
Epidemiology and Public Health: How They Relate to Each Other and the Concept of One Health
Individual Health and Population Health
The Articulation Between Individual and Population Level
Biomedical and Biopsychosocial Models of Health: Individual, Environmental and Social Determinants of Health
From Precision Medicine to Precision Public Health
Epidemiology and Public Health in the Digital Era: Prerequisites
A Ubiquitous Digitization
The Evolutions of the Regulatory Framework on Personal Data
Connected Devices and Equipment Rates
Digital and E-health Literacy
Towards a Real Life Use of AI in Epidemiology and Public Health: Some First Examples
No Data Means No Artificial Intelligence: A Few Words About Data Federation and “New” Types of Data
Citizens and Patients as Producers, Actor and Manager of Their Own Health
At the Population Level, Health Surveillance Systems and AI
Between the Individual and the Population, Healthcare Systems: Learning Healthcare Systems (LHS)
What Contributions Could Be Expected from AI in Epidemiology and Public Health in the Context of a Pandemic?
What Is Due to the Use of Non-classical Data Sources, and to the Comparison or Cross-Checking Between Sources
The Real-Time or Refreshable Nature of the Information
New Types of Analysis, in Particular on Massive, Incomplete or Even Poorly Balanced Data
Aid in the Search for Causality or Networks of Causality
Evaluation Methods Based on Observation or Quasi-Experimental, as for Virtual Clinical Trials
The Augmented Expert in Public Health
Some Specificities and Examples in the Context of the Pandemic: Epidemic Modeling, Public Health and Counting Deaths
Epidemic Modeling: How to Retrieve Accurate and Timely Data to Feed a Model
Counting Deaths in the Context of a Health Crisis Based on Information Systems and AI
Challenges of Counting Death, the Case of the French System
Certification of Deaths and Collection of Death Certificates, Specificity to Emerging Diseases Such as COVID
Issues in Coding Causes of Death, Especially in Emerging Diseases
AI’s Place in Coding Deaths on Emerging Causes/Diseases, COVID Contextual Example
To Take Away
AI at the Service of Epidemiology and Public Health in the Context of the Covid19 Pandemic
Bibliographic Search
A Typology of AI Use at the Service of Epidemiology and Public Health in the Context of the Pandemic
Outbreak Monitoring
Epidemiologic Outcomes and Characteristics Discovering
Social Control and Monitoring
Assisted/Augmented Scientific Research and Knowledge Sharing
Healthcare Resources Adaptation and Optimization
Social, Economic and Governmental Measures Assessment
Infodemics
What Performance of AI in the Uses Identified in Epidemiology and Public Health?
Outbreak Monitoring
Epidemiologic Outcomes and Characteristics Discovering
Social Control and Monitoring
Assisted/Augmented Scientific Research and Knowledge Sharing
Healthcare Resources Adaptation and Optimization
Social, Economic and Government Measures Assessment
Infodemics
Beyond Performance, What About the Degree of Maturity of Published AI Algorithms?
Outbreak Monitoring
Epidemiologic Outcomes and Characteristics Discovering
Social Control and Monitoring
Assisted/Augmented Scientific Research and Knowledge Sharing
Healthcare Resources Adaptation and Optimization
Social, Economic and Government Measures Assessment
Infodemics
Two Years of Pandemic: Lessons for Epidemiology and the Place of AI
Many AI Applications for Epidemiology and Public Health in the Context of the Pandemic: Yet Still Evidence for Their Reliability and Usefulness to Bring
“Lancetgate”: A Lesson About Identified Risks of Massive Data Collection and Reuse and Its Consequences on Public Health Decisions
What AI Has to Learn from Epidemiology and Public Health?
Predicting vs. Explaining: Is It Reconcilable? Is Explainable AI Necessary in Epidemiology?
The Status of the Whistleblower in the Case of Emerging Diseases: The Hybrids of Simondon and Latour
Towards a Potentially More Actionable and Precise Public Health: The Challenges of Regulation, Ethics and Ecology on an International Level
Benefice/Risk Balance of the Use of AI in the Context of Population Management
The Need for a Cautious Analysis of the Real Cost of AI Use in Health
References
Index