Cover Image

Federated learning for digital healthcare systems /

Federated Learning for Digital Healthcare Systems critically examines the key factors that contribute to the problem of applying machine learning in healthcare systems and investigates how federated learning can be employed to address the problem.

Bibliographic Details
Corporate Author: ScienceDirect (Online service)
Other Authors: Imoize, Agbotiname Lucky
Format: eBook
Language:English
Published: London ; San Diego, CA : Academic Press, an imprint of Elsevier, [2024]
Subjects:
Machine learning.
Artificial intelligence > Medical applications.
Machine Learning
Apprentissage automatique.
Intelligence artificielle > Applications en médecine.
Electronic books.
Online Access:Connect to the full text of this electronic book
Table of Contents:
  • Front Cover
  • Federated Learning for Digital Healthcare Systems
  • Copyright Page
  • Contents
  • List of contributors
  • Preface
  • 1 Digital healthcare systems in a federated learning perspective
  • 1.1 Introduction
  • 1.1.1 Chapter organization
  • 1.2 Federated learning
  • 1.2.1 Statistical challenges of federated learning
  • 1.2.1.1 Consensus solution
  • 1.2.1.2 Pluralistic solution
  • 1.2.2 Security and privacy
  • 1.2.2.1 Differential privacy
  • 1.2.2.2 Secure multiparty computation
  • 1.2.3 Federated learning communication efficiency
  • 1.2.4 User selection
  • 1.2.5 Model compression
  • 1.2.6 Updates reduction
  • 1.2.7 Peer-to-peer learning
  • 1.3 Federated learning data processing: as still, stream, and multistream
  • 1.3.1 Data processing at rest
  • 1.3.1.1 Data collection
  • 1.3.1.2 Batch or interval-based processing
  • 1.3.1.2.1 Data preprocessing
  • 1.3.1.2.2 Model training
  • 1.3.1.2.3 Model aggregation
  • 1.3.1.2.4 Model improvement
  • 1.3.1.3 Merits of data processing at rest
  • 1.3.1.4 Considerations on data processing at rest
  • 1.3.2 Data processing as a stream
  • 1.3.2.1 Data ingestion
  • 1.3.2.2 Stream processing
  • 1.3.2.3 Advantages of data processing as a stream
  • 1.3.2.4 Considerations
  • 1.3.3 Multistream data processing
  • 1.3.3.1 Data streams
  • 1.3.3.2 Stream processing
  • 1.3.3.3 Advantages of multistream data processing
  • 1.3.3.4 Considerations of multistream data processing
  • 1.4 Applications
  • 1.5 Healthcare
  • 1.6 Other applications
  • 1.7 Federated learning in healthcare ecosystem
  • 1.7.1 Privacy preservation
  • 1.7.2 Healthcare data fragmentation and silos
  • 1.7.3 Data heterogeneity
  • 1.7.4 Regulatory compliance
  • 1.7.5 Healthcare resource efficiency
  • 1.7.6 Bias mitigation
  • 1.7.7 Collaborative learning and healthcare research
  • 1.7.8 Real-time medical monitoring and edge computing.
  • 1.8 Federated learning open research questions
  • 1.8.1 Model precision
  • 1.8.2 Personalization
  • 1.8.3 Incentive mechanisms
  • 1.8.4 Expert knowledge incorporation
  • 1.8.5 Data quality
  • 1.9 Federated learning challenges in smart healthcare
  • 1.9.1 Data privacy and security
  • 1.9.2 Heterogeneity of data
  • 1.9.3 Data quality
  • 1.9.4 Communication overhead
  • 1.9.5 Regulatory compliance
  • 1.9.6 Resource constraints
  • 1.10 Future federated learning trends
  • 1.10.1 Increased use of wearable technology
  • 1.10.2 Personalized medicine
  • 1.10.3 Improved patient outcomes
  • 1.10.4 Increased data privacy
  • 1.10.5 Remote patient monitoring
  • 1.10.6 Collaborative healthcare
  • 1.10.7 Real-time decision-making
  • 1.10.8 Improved drug discovery
  • 1.10.9 Predictive analytics
  • 1.10.10 Improved clinical trials
  • 1.10.11 Enhanced medical imaging
  • 1.10.12 Continuous learning
  • 1.10.13 Improved resource allocation
  • 1.10.14 Increased patient engagement
  • 1.10.15 Improved public health
  • 1.11 Conclusion
  • Abbreviations
  • References
  • Further reading
  • 2 Architecture and design choices for federated learning in modern digital healthcare systems
  • 2.1 Introduction
  • 2.1.1 Key contributions
  • 2.1.2 Chapter organization
  • 2.2 Dataspaces and health domain
  • 2.2.1 State-of-the-art and current practices
  • 2.2.2 Impact on machine learning
  • 2.2.3 European digital age and development of secure (health) dataspaces
  • 2.2.4 Potential
  • 2.2.5 Challenges
  • 2.3 Proposed approach
  • 2.3.1 Data life cycle
  • 2.3.2 Data generation and interoperability
  • 2.3.3 Data exposure
  • 2.3.4 Data discovery
  • 2.3.5 Data usage
  • 2.4 Dataspaces and participation in ecosystems
  • 2.4.1 Identity governance
  • 2.4.2 Consent management
  • 2.4.3 Trust establishment
  • 2.4.4 Protection and assurance
  • 2.4.4.1 Privacy-preserving technologies.
  • 2.4.4.1.1. Agent-based approach
  • 2.4.4.1.2 Machine learning-based approach
  • 2.4.4.2 Secure multiparty computation
  • 2.4.4.3 Differential privacy
  • 2.4.4.4 Data protection and traceability
  • 2.5 Lessons learned: conclusions and future scope
  • Acknowledgment
  • References
  • 3 Curation of federated patient data: a proposed landscape for the African Health Data Space
  • 3.1 Introduction
  • 3.2 Background
  • 3.2.1 Inventory on open science and findable, accessible, interoperable, and reusable data initiatives in Africa
  • 3.2.2 Relevance of federated learning approaches for digital healthcare
  • 3.3 Conceptual framework
  • 3.4 Methodology
  • 3.4.1 Research approach
  • 3.4.2 Virus outbreak data network: establishing a quality data production pipeline in residence
  • 3.4.3 Installation of a federated data network
  • 3.5 Results: use cases
  • 3.5.1 Pandemic early warning
  • 3.5.2 Integration of data on incidence of COVID-19
  • 3.5.3 Regional dashboard for monitoring at health-bureau level in Tigray
  • 3.5.4 Syphilis cases in Ayder Referral Hospital
  • 3.5.5 Scaling up health system implementation research: the Saving Little Lives project
  • 3.5.6 Retrospective FAIRification to identify drivers of perinatal care outcomes in Kenya and Tanzania (MomCare)
  • 3.5.7 FAIR data curation of clinical research data on vaccines
  • 3.5.8 Integration of cardio-related patient-generated health data
  • 3.5.9 Interoperability across sectors: data on human trafficking
  • 3.6 Landscape for an African Health Data Space
  • 3.7 Discussion
  • 3.8 Conclusion
  • 3.9 Ethical clearance
  • Acknowledgments
  • References
  • 4 Recent advances in federated learning for digital healthcare systems
  • 4.1 Introduction
  • 4.1.1 Key contributions of the chapter
  • 4.1.2 Chapter organization
  • 4.2 Related works.
  • 4.3 Perceptions of federated digital platforms and their use in healthcare
  • 4.3.1 Use of federated learning in medical image processing
  • 4.3.2 Use of federated learning in Internet of Things-based smart healthcare applications
  • 4.4 Privacy preservation, security, and ethical needs
  • 4.5 Privacy and security Needs
  • 4.5.1 Data privacy
  • 4.5.2 Model poisoning attacks
  • 4.5.3 Differential privacy
  • 4.5.4 Secure model aggregation
  • 4.5.5 Data minimization
  • 4.5.6 Secure and encrypted communication
  • 4.6 Ethical needs
  • 4.7 Role of federated learning in future digital Healthcare 5.0
  • 4.8 Federated learning and blockchain for healthcare
  • 4.9 Federated learning for collaborative robotics in healthcare
  • 4.10 Federated learning for integration with 6G in healthcare
  • 4.11 Conclusion
  • Acknowledgment
  • References
  • 5 Performance evaluation of federated learning algorithms using breast cancer dataset
  • 5.1 Introduction
  • 5.2 Related works/literature reviews
  • 5.3 Federated learning
  • 5.4 Inspiration for federated learning
  • 5.5 Federated learning algorithm
  • 5.5.1 Federated averaging
  • 5.5.2 Federated match averaging
  • 5.6 Materials and methods/methodology/design
  • 5.6.1 Study dataset
  • 5.7 Transfer learning
  • 5.8 Pretrained classifiers: a short overview
  • 5.9 Visual geometry group network
  • 5.9.1 Visual geometry group-16
  • 5.9.2 Visual geometry group network-19
  • 5.10 Convolutional neural network model for federated learning
  • 5.11 Model evaluation
  • 5.12 Results and analysis
  • 5.13 Conclusion and future scope
  • References
  • 6 Taxonomy for federated learning in digital healthcare systems
  • 6.1 Introduction
  • 6.1.1 Key contributions of the chapter
  • 6.1.2 Chapter organization
  • 6.2 Related works
  • 6.3 Fundamentals of federated learning
  • 6.3.1 Federated learning.
  • 6.3.2 System architectures of federated learning used in healthcare systems
  • 6.3.3 Existing frameworks of federated learning in healthcare systems
  • 6.3.4 Underlying technologies in federated learning in healthcare systems
  • 6.3.5 Challenges in implementing federated learning
  • 6.3.6 Privacy-preservation methods in federated learning
  • 6.4 Taxonomy of critical aspects of federated learning in healthcare systems
  • 6.4.1 Centralized machine learning versus federated learning
  • 6.4.2 Cloud computing versus edge computing
  • 6.4.3 Assumptions and characterization of federated learning
  • 6.4.3.1 Assumptions
  • 6.4.3.1.1 Availability of sufficient data
  • 6.4.3.1.2 Data privacy and security
  • 6.4.3.1.3 Data heterogeneity and quality
  • 6.4.3.1.4 Trust and cooperation
  • 6.4.3.1.5 Regulatory and ethical compliance
  • 6.4.3.2 Characterization
  • 6.4.3.2.1 Decentralized data ownership
  • 6.4.3.2.2 Privacy-preserving collaboration
  • 6.4.3.2.3 Data diversity and representativeness
  • 6.4.3.2.4 Resource efficiency and scalability
  • 6.4.3.2.5 Transferability and knowledge sharing
  • 6.4.3.2.6 Collaboration with regulatory frameworks
  • 6.4.4 Applications and trending use cases of federated learning in healthcare systems
  • 6.4.4.1 Personalized medicine
  • 6.4.4.2 Disease surveillance and early detection
  • 6.4.4.3 Medical imaging analysis
  • 6.4.4.4 Drug discovery and development
  • 6.4.4.5 Remote monitoring and telemedicine
  • 6.4.4.6 Clinical decision support systems
  • 6.4.5 Classification and clustering of federated learning in digital healthcare systems
  • 6.4.5.1 Approach or data type
  • 6.4.5.2 Federated learning architecture
  • 6.4.5.3 Application type
  • 6.4.5.4 Data modality
  • 6.4.5.5 Privacy level
  • 6.4.5.6 Task
  • 6.4.5.7 Learning mode
  • 6.4.6 Future directions and challenges in federated learning within digital healthcare systems.