Federated Learning for Healthcare: A Comprehensive Review

Feature extraction and selection from medical data are the basis of radiomics and image biomarker discovery for various architectures, including convolutional neural networks (CNNs). We herein describe the typical radiomics steps and the components of a CNN for both deep feature extraction and end-to-end approaches. We discuss the curse of dimensionality, along with dimensionality reduction techniques. Despite the outstanding performance of deep learning (DL) approaches, the use of handcrafted features instead of deep learned features needs to be considered for each specific study. Dataset size is a key factor: large-scale datasets with low sample diversity could lead to overfitting; limited sample sizes can provide unstable models. The dataset must be representative of all the “facets” of the clinical phenomenon/disease investigated. The access to high-performance computational resources from graphics processing units is another key factor, especially for the training phase of deep architectures. The advantages of multi-institutional federated/collaborative learning are described. When large language models are used, high stability is needed to avoid catastrophic forgetting in complex domain-specific tasks. We highlight that non-DL approaches provide model explainability superior to that provided by DL approaches. To implement explainability, the need for explainable AI arises, also through post hoc mechanisms. Relevance statement This work aims to provide the key concepts for processing the imaging features to extract reliable and robust image biomarkers. Key Points The key concepts for processing the imaging features to extract reliable and robust image biomarkers are provided. The main differences between radiomics and representation learning approaches are highlighted. The advantages and disadvantages of handcrafted versus learned features are given without losing sight of the clinical purpose of artificial intelligence models. Graphical Abstract

show abstract

Decades in the Making: The Evolution of Digital Health Research Infrastructure Through Synthetic Data, Common Data Models, and Federated Learning

Austin,

Lobo,

Samadbeik

et al. 2024

J Med Internet Res

View full text Add to dashboard Cite

Traditionally, medical research is based on randomized controlled trials (RCTs) for interventions such as drugs and operative procedures. However, increasingly, there is a need for health research to evolve. RCTs are expensive to run, are generally formulated with a single research question in mind, and analyze a limited dataset for a restricted period. Progressively, health decision makers are focusing on real-world data (RWD) to deliver large-scale longitudinal insights that are actionable. RWD are collected as part of routine care in real time using digital health infrastructure. For example, understanding the effectiveness of an intervention could be enhanced by combining evidence from RCTs with RWD, providing insights into long-term outcomes in real-life situations. Clinicians and researchers struggle in the digital era to harness RWD for digital health research in an efficient and ethically and morally appropriate manner. This struggle encompasses challenges such as ensuring data quality, integrating diverse sources, establishing governance policies, ensuring regulatory compliance, developing analytical capabilities, and translating insights into actionable strategies. The same way that drug trials require infrastructure to support their conduct, digital health also necessitates new and disruptive research data infrastructure. Novel methods such as common data models, federated learning, and synthetic data generation are emerging to enhance the utility of research using RWD, which are often siloed across health systems. A continued focus on data privacy and ethical compliance remains. The past 25 years have seen a notable shift from an emphasis on RCTs as the only source of practice-guiding clinical evidence to the inclusion of modern-day methods harnessing RWD. This paper describes the evolution of synthetic data, common data models, and federated learning supported by strong cross-sector collaboration to support digital health research. Lessons learned are offered as a model for other jurisdictions with similar RWD infrastructure requirements.

show abstract

Federated Learning for Healthcare: A Comprehensive Review

Abstract: This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

Cited by 8 publications

References 29 publications

KL-FedDis: A federated learning approach with distribution information sharing using Kullback-Leibler divergence for non-IID data

KL-FedDis: A federated learning approach with distribution information sharing using Kullback-Leibler divergence for non-IID data

Image biomarkers and explainable AI: handcrafted features versus deep learned features

Decades in the Making: The Evolution of Digital Health Research Infrastructure Through Synthetic Data, Common Data Models, and Federated Learning

Contact Info

Product

Resources

About