Continuous and real-time health monitoring plays a crucial role in tracking disease progression and enabling timely interventions through early alerts. Wearable edge devices have can perform real-time health monitoring, but must operate for extended periods on small batteries. Their computational efficiency is therefore key.
Incremental inference and learning efforts pursue efficiency at the algorithmic level. Our approach adapts state of the art workflows by automatically deriving incremental versions. In these, only a subset of clinical features are computed and processed given a window of input data, i.e. the ones required to reach a confident decision. In this way, computational efforts can be dynamically adapted, employing large features sets (which require a lot of energy to compute) only when necessary.
![Workflow for the transformation of monolithic health monitoring application into incremental ones, from [1]](https://eslweb.epfl.ch/masters/img/20241003incremental.png)
Workflow for the transformation of monolithic health monitoring application into incremental ones, from [1]
Related Publications
| ACE: Automated optimization towards iterative Classification in Edge health monitors | ||||
| Yuxuan Wang; Lara Orlandic; Simone Machetti; Giovanni Ansaloni; David Atienza | ||||
| 2024-09-25 | IntIEEE Transactions on Biomedical Circuits and Systems | Early Access | ||
| INCLASS: Incremental Classification Strategy for Self-Aware Epileptic Seizure Detection | ||||
| Ferretti, Lorenzo; Ansaloni, Giovanni; Marquis, Renaud; Teijeiro, Tomas; Ryvlin, Philippe; Atienza, David; Pozzi, Laura | ||||
| 2022-03-23 | Design Automation and Test in Europe Conference (DATE2022) | |||
| A Self-Aware Epilepsy Monitoring System for Real-Time Epileptic Seizure Detection | ||||
| Forooghifar, Farnaz ; Aminifar, Amir ; Cammoun, Leila ; Atienza Alonso, David ; Wisniewski, Ilona ; Ciumas, Carolina ; Ryvlin, Philippe | ||||
| 2019-08-08 | Mobile Networks and Applications |  |  | |