Decentralized Epidemic Tracking using Federated Learning and Edge AI on Consumer Wearables

This paper has proposed a decentralized epidemic-tracking framework that combines federated learning and edge AI on consumer wearables. Personalized on-device detection, hierarchical privacy-protected aggregation, federated analytics with k-anonymous geo-cells, energy-aware participation, an asynchronous permissioned audit trail, federated unlearning, and a pathogen-agnostic cold-start strategy together target the constraints that the wearable FL literature identifies as unresolved: privacy and security, energy, convergence, trust, and regulation. Six learning approaches were compared conceptually and a multi-objective evaluation design was defined. Because no dataset was collected for this study, the work is presented explicitly as a proposed framework and experimental design rather than a report of measured results, and all illustrative figures and tables are labelled accordingly. Moving from concept to a field-validated public-health tool will require prospective pilots, rigorous privacy and fairness evaluation, and regulatory engagement.

[1] G. Ranjbaran, S. Consoli, G. Leoni, D. Reforgiato Recupero, and C. K. Roy, "Federated artificial intelligence approaches for wearables and health devices," Smart Health, vol. 40, p. 100663, 2026, doi: 10.1016/j.smhl.2026.100663. [2]   T. Mishra et al., "Pre-symptomatic detection of COVID-19 from smartwatch data," Nature Biomedical Engineering, vol. 4, pp. 1208–1220, 2020. [3]  G. Quer et al., "Wearable sensor data and self-reported symptoms for COVID-19 detection," Nature Medicine, vol. 27, pp. 73–77, 2021. [4]  A. B. Grzesiak et al., "Assessment of the feasibility of using noninvasive wearable biometric monitoring sensors to detect influenza and the common cold before symptom onset," JAMA Network Open, vol. 4, no. 9, e2128534, 2021. [5] F. Cacheda, V. Carneiro, M. A. Álvarez, and M. F. López-Vizcaíno, "Early detection of infectious diseases using wearable sensor data through personalized baseline deviation modelling," Digital Health, vol. 12, 2026, doi: 10.1177/20552076261477919. [6]  H. B. McMahan, E. Moore, D. Ramage, S. Hampson, and B. Agüera y Arcas, "Communication-efficient learning of deep networks from decentralized data," in Proc. AISTATS, 2017. [7]  K. Bonawitz et al., "Practical secure aggregation for privacy-preserving machine learning," in Proc. ACM CCS, 2017. [8]   C. Dwork and A. Roth, "The algorithmic foundations of differential privacy," Foundations and Trends in Theoretical Computer Science, vol. 9, nos. 3–4, pp. 211–407, 2014. [9]  T. Li, A. K. Sahu, M. Zaheer, M. Sanjabi, A. Talwalkar, and V. Smith, "Federated optimization in heterogeneous networks," in Proc. MLSys, 2020. [10]  L. Bourtoule et al., "Machine unlearning," in Proc. IEEE Symposium on Security and Privacy, 2021.