Federated learning and Indigenous genomic data sovereignty.

Area of Science:

• Genomics
• Bioinformatics
• Health Equity

Background:

• Indigenous peoples are significantly under-represented in current genomic datasets.
• This under-representation limits the accuracy and applicability of precision health initiatives for these communities.
• Existing data sharing models can conflict with Indigenous rights to data governance.

Purpose of the Study:

• To highlight the challenges posed by under-representation of Indigenous peoples in genomic data.
• To explore federated learning as a potential solution for secure and consented data sharing.
• To promote equitable advancements in precision health for Indigenous populations.

Main Methods:

• Review of current genomic data sharing practices and their impact on Indigenous communities.
• Analysis of federated learning principles and their applicability to genomic data.
• Examination of ethical considerations for community-consented data use.

Main Results:

• Under-representation in genomic datasets leads to biased machine learning models in precision health.
• Federated learning enables model training on decentralized data without direct data sharing.
• This approach can uphold Indigenous data sovereignty and facilitate collaborative research.

Conclusions:

• Addressing the under-representation of Indigenous peoples in genomic data is crucial for equitable precision health.
• Federated learning presents a promising, privacy-preserving method for Indigenous genomic data sharing.
• Implementing federated learning requires strong community engagement and adherence to Indigenous data governance principles.