Towards fairness-aware and privacy-preserving enhanced collaborative learning for healthcare

Area of Science:

• Artificial Intelligence in Healthcare
• Machine Learning Ethics
• Health Informatics

Background:

• AI integration in healthcare raises privacy and fairness concerns.
• Federated Learning (FL) enables collaborative model training on distributed data without direct patient record access, enhancing privacy.
• Disparities in computing resources among healthcare institutions create unfairness in FL participation and model performance.

Purpose of the Study:

• To address the performance gap between resource-limited and resource-rich institutions in Federated Learning.
• To propose a novel resource-adaptive framework for equitable collaborative learning in healthcare.
• To ensure fair participation and enhance AI-driven healthcare outcomes for all institutions.

Main Methods:

• Development of a resource-adaptive framework for collaborative learning.
• Dynamic adjustment of the framework to accommodate varying computational capacities of healthcare institutions.
• Implementation of strategies to ensure fair participation and mitigate performance disparities.

Main Results:

• The proposed framework enhances model accuracy and safeguards patient privacy.
• It ensures equitable access to trustworthy and efficient AI-driven healthcare solutions.
• It dynamically adjusts to varying computational capacities, promoting fair participation.

Conclusions:

• The resource-adaptive framework effectively addresses fairness and performance gaps in healthcare Federated Learning.
• It promotes equitable access to advanced AI healthcare solutions, reducing disparities.
• This approach supports trustworthy and efficient AI deployment in diverse healthcare settings.