Energy-Efficient TinyML Architectures for AI-Driven Wearable Health Monitoring
Keywords:
TinyML, wearable health monitoring, energy-efficient deep learning, edge AI, neural architecture search, federated learning, fairness, sustainability, system-level designAbstract
The integration of artificial intelligence into wearable health monitoring systems promises transformative advances in continuous physiological assessment, early disease detection, and personalized intervention. However, the severe energy constraints of battery-powered wearable devices impose fundamental limits on the complexity of onboard machine learning models. This paper presents a comprehensive systems-level analysis of energy-efficient TinyML architectures designed specifically for AI-driven wearable health monitoring. We examine structural trade-offs between model accuracy, computational latency, memory footprint, and energy consumption across diverse neural network design paradigms, including depthwise separable convolutions, neural architecture search, quantization-aware training, and knowledge distillation. Beyond algorithmic considerations, we address the broader infrastructure necessary for sustainable deployment, including heterogeneous processing hardware, on-device inference scheduling, and federated learning governance. The paper also critically evaluates robustness, fairness, and ethical implications of TinyML-based health decisions, emphasizing the need for rigorous validation across diverse populations and clinical contexts. Policy challenges such as data privacy, algorithmic accountability, and regulatory compliance are analyzed in the context of edge-based health analytics. By synthesizing recent advances in ultra-low-power machine learning with socio-technical requirements, this work provides a roadmap for designing trustworthy, scalable, and energy-sustainable wearable health monitoring systems.
References
1. Howard, A. G., Zhu, M., Chen, B., Kalenichenko, D., Wang, W., Weyand, T., ... & Adam, H. (2017). MobileNets: Efficient convolutional neural networks for mobile vision applications. arXiv preprint arXiv:1704.04861.
2. Sandler, M., Howard, A., Zhu, M., Zhmoginov, A., & Chen, L. C. (2018). MobileNetV2: Inverted residuals and linear bottlenecks. Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, 4510–4520.
3. Zoph, B., & Le, Q. V. (2017). Neural architecture search with reinforcement learning. arXiv preprint arXiv:1611.01578.
4. Tan, M., Chen, B., Pang, R., Vasudevan, V., Sandler, M., Howard, A., & Le, Q. V. (2019). MnasNet: Platform-aware neural architecture search for mobile. Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, 2820–2828.
5. Chen, J., Zhang, Y., Xu, Z., & Li, Q. (2020). MCUNet: Tiny deep learning on IoT devices. Advances in Neural Information Processing Systems, 33, 11711–11722.
6. Jacob, B., Kligys, S., Chen, B., Zhu, M., Tang, M., Howard, A., ... & Adam, H. (2018). Quantization and training of neural networks for efficient integer-arithmetic-only inference. Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, 2704–2713.
7. Hubara, I., Courbariaux, M., Soudry, D., El-Yaniv, R., & Bengio, Y. (2017). Quantized neural networks: Training neural networks with low precision weights and activations. Journal of Machine Learning Research, 18(1), 6869–6898.
8. Rastegari, M., Ordonez, V., Redmon, J., & Farhadi, A. (2016). XNOR-Net: ImageNet classification using binary convolutional neural networks. European Conference on Computer Vision, 525–542.
9. Hinton, G., Vinyals, O., & Dean, J. (2015). Distilling the knowledge in a neural network. arXiv preprint arXiv:1503.02531.
10. Polino, A., Pascanu, R., & Alistarh, D. (2018). Model compression via distillation and quantization. arXiv preprint arXiv:1802.05668.
11. Krishnan, S., Chai, E. G., & Sim, D. Y. (2021). MEMS-based energy harvesting for wearable devices: A review. Sensors, 21(9), 3012.
12. McMahan, B., Moore, E., Ramage, D., Hampson, S., & y Arcas, B. A. (2017). Communication-efficient learning of deep networks from decentralized data. Artificial Intelligence and Statistics, 1273–1282.
13. Ren, J., Yu, G., He, Y., & Li, G. Y. (2020). Collaborative cloud and edge computing for latency minimization in mobile networks. IEEE Transactions on Communications, 68(10), 6385–6399.
14. Wu, C., Li, J., Liu, Y., & He, X. (2020). TinyTL: Reduce memory, not parameters for efficient on-device learning. Advances in Neural Information Processing Systems, 33, 11285–11297.
15. Sze, V., Chen, Y. H., Yang, T. J., & Emer, J. S. (2017). Efficient processing of deep neural networks: A tutorial and survey. Proceedings of the IEEE, 105(12), 2295–2329.
16. Dai, S., Li, J., & Zhang, W. (2022). Energy-adaptive TinyML for sustainable wearable health monitoring. ACM Transactions on Embedded Computing Systems, 21(4), 1–24.
17. Zhang, C., Patras, P., & Haddadi, H. (2019). Deep learning in mobile and wireless networking: A survey. IEEE Communications Surveys & Tutorials, 21(3), 2224–2287.
18. Tan, M., & Le, Q. V. (2019). EfficientNet: Rethinking model scaling for convolutional neural networks. Proceedings of the 36th International Conference on Machine Learning (ICML), 6105–6114.
19. Rothchild, D., Panda, A., Ullah, E., Ivkin, N., Stoica, I., Braverman, V., & Arora, R. (2020). FetchSGD: Communication-efficient federated learning with sketching. Proceedings of the 37th International Conference on Machine Learning (ICML), 8271–8282.
20. Gu, T., Dolan-Gavitt, B., & Garg, S. (2017). BadNets: Identifying vulnerabilities in the machine learning model supply chain. arXiv preprint arXiv:1708.06733.
Downloads
Published
Issue
Section
License
Copyright (c) 2026 Journal of Advanced Artificial Intelligence Research
This work is licensed under a Creative Commons Attribution 4.0 International License.
This article is published under the Creative Commons Attribution 4.0 International License (CC BY 4.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
