Hantavirus genomic surveillance is limited by the distribution of sequence data, non-IID source heterogeneity, and constrained expert-review capacity. We propose HantaWatch, a federated learning framework that enables laboratories and surveillance sites to collaboratively train sequence-based models without sharing raw data.
HantaWatch integrates several core components:
- k-mer feature extraction: Extracting important features from sequence data for model training.
- Source-aware federated client construction: Building effective clients based on the characteristics of data sources.
- Adaptive DU-FedProx optimization: Utilizing advanced optimization algorithms to improve model convergence speed and stability.
- Surveillance-specific model selection: Selecting appropriate models based on task requirements.
- Prediction-only triage: Enabling rapid decision support.
Experiments on binary and multi-class tasks show that HantaWatch supports high-risk screening, outbreak-associated prediction, clade classification, and clinical-syndrome categorization, while balancing predictive performance, false-negative risk, and update stability. The framework converts model output into risk scores, confidence estimates, uncertainty flags, and ranked expert-review priorities, providing a practical federated decision-support layer for decentralized hantavirus surveillance, supporting expert prioritization without replacing laboratory or public health interpretation.
Blogger's Review: The innovation of HantaWatch lies in its application of federated learning, enhancing the efficiency and accuracy of hantavirus surveillance while protecting data privacy. The design of this framework fully considers the heterogeneity of different data sources, offering strong support for real-time decision-making in public health. Its potential future applications are worth monitoring.