Page 48 - AIH-2-3

Artificial Intelligence in Health Omnichain FL in healthcare AI

federated learning model can seamlessly access and Omnichain, the federated learning framework may
aggregate these diverse data sources – while respecting decrease redundant deployments, thus enabling support
privacy and permission constraints – thus enhancing for multiple blockchains directly from the Omnichain. This
the model’s generalization capabilities and training design enhances the efficiency of the AI training paradigm
quality. In parallel, the concept of a taxonomy driven by that combines federated learning with blockchain in
practical services can be applied here. By approaching healthcare. However, it is important to note that the
the discussion from the perspective of concrete services, Omnichain is still in its nascent developmental stage, and
we can deduce the unique advantages that Omnichain its speed and cost remain significant constraints on its
offers, thereby realizing synergy between a fully integration with federated learning. This presents a crucial
integrated blockchain framework and federated learning area for future research and attention.
across diverse domains, such as health data circulation
and management, privacy and security, token-based Acknowledgments
incentive mechanisms, collaborative orchestration, and We are deeply grateful to the Duke University Writing
regulatory oversight. This approach ultimately advances Studio for tremendous assistance in polishing the academic
the real-world adoption of the hybrid solution. For language of our work.
example, in the realm of health data circulation and
management, indexing services built on Omnichain Funding
can standardize data drawn from multiple blockchains,
thereby streamlining cross-chain data flow and providing None.
high-quality datasets for federated learning. Conflict of interest
However, Omnichain remains undeveloped. Before The authors declare that they have no competing interests.
its development, cross-chain bridges represented a
bold yet flawed endeavor. Users were required to lock Author contributions
assets on the source chain and incur gas fees to receive
corresponding wrapped assets on the target chain, Conceptualization: Dongfang Wu
thereby creating liquidity challenges. In addition, Writing–original draft: Dongfang Wu
10
cross-chain bridges predominantly focused on value Writing–review & editing: Yichen Wang
transfer rather than imperative information transfer, Ethics approval and consent to participate
which is essential for training AI. In contrast, Omnichain
emerges as a novel paradigm extending from cross- Not applicable.
chain bridges, utilizing universal smart contracts to
create an infrastructure that spans multiple chains and Consent for publication
can be directly deployed across various blockchains, Not applicable.
including Ethereum Virtual Machine – compatible and
other mainstream platforms, thereby mitigating the silo Availability of data
effect of disparate blockchains. Nevertheless, the high- Not applicable.
throughput requirements of federated learning in the
training process remain in tension with the existing References
gas pricing mechanisms of Omnichain, and the block
generation speed of Omnichain further constrains 1. Truong N, Sun K, Wang S, Guitton F, Guo Y. Privacy
11
the deployment of federated learning solutions. This preservation in federated learning: An insightful survey from
the GDPR perspective. Comput Secur. 2021;110:102402.
necessitates that Omnichain, much like various Layer-2
solutions of Ethereum, progresses toward achieving low- doi: 10.1016/j.cose.2021.102402
cost and rapid transaction capabilities. 2. Zhan Y, Li P, Qu Z, Zeng D, Guo S. A learning-based
incentive mechanism for federated learning. IEEE Internet
4. Conclusion Things J. 2020;7:6360-6368.

This article proposes the integration of Omnichain concept doi: 10.1109/JIOT.2020.2967772
into the existing frameworks of federated learning and 3. Mammen PM. Federated learning: Opportunities and
blockchain, with the objective of minimizing performance challenges. arXiv [Preprint]. 2021.
degradation while maintaining privacy protection
during AI training in the healthcare sector. By deploying doi: 10.48550/arXiv.2101.05428
smart contracts tailored for diverse blockchains on the 4. Nguyen DC, Ding M, Pham QV, et al. Federated learning

Volume 2 Issue 3 (2025) 42 doi: 10.36922/aih.5753