Assessing body position during sleep using FSR sensors and machine learning algorithms
- This study investigates the application of Force Sensing Resistor (FSR) sensors and machine learning algorithms for non-invasive body position monitoring during sleep. Although reliable, traditional methods like Polysomnography (PSG) are invasive and unsuited for extended home-based monitoring. Our approach utilizes FSR sensors placed beneath the mattress to detect body positions effectively. We employed machine learning techniques, specifically Random Forest (RF), K-Nearest Neighbors (KNN), and XGBoost algorithms, to analyze the sensor data. The models were trained and tested using data from a controlled study with 15 subjects assuming various sleep positions. The performance of these models was evaluated based on accuracy and confusion matrices. The results indicate XGBoost as the most effective model for this application, followed by RF and KNN, offering promising avenues for home-based sleep monitoring systems.
| Author of HS Reutlingen | Martínez Madrid, Natividad |
|---|---|
| URL: | http://www.dii.univpm.it/MAES-2023 |
| Erschienen in: | Models and applications for embedded systems |
| Publisher: | Università Politecnica delle Marche |
| Place of publication: | Ancona |
| Editor: | Massimo Conti, Simone Orcioni |
| Document Type: | Book chapter |
| Language: | English |
| Publication year: | 2024 |
| Page Number: | 4 |
| First Page: | 19 |
| Last Page: | 22 |
| DDC classes: | 610 Medizin, Gesundheit |
| Open access?: | Nein |
| Licence (German): |  In Copyright - Urheberrechtlich geschützt |
