Please use this identifier to cite or link to this item:
http://hdl.handle.net/10609/127707
Title: | Detección de caídas con reloj inteligente (smartwatch) para personas mayores |
Author: | Tienda Tejedo, Arnau |
Tutor: | Pérez Álvarez, Susana |
Abstract: | The percentage of people over 65 years old in Spain increases year after year. A significant part of this group lives alone in their homes, a fact that makes families worry about their health condition as they do not receive continuous feedback from, for example, a caregiver. The purpose of this project is to analyze the viability of creating an smartwatch with two functions: detection of falls, and analysis of atrial fibrillation, the most common type of arrhythmia and the one that causes more strokes. Regarding accelerometer data, it has been possible to find enough labeled data, that is, accelerometer readings that indicate the activity being performed. In addition, it is feasible to create your own data from the smartwatch by performing actions and capturing them, as to get a good model that classifies falls. We have trained 4 supervised learning methods, and we have managed to obtain a fall detection with a very high sensitivity, classifying correctly up to 96% of the falls. On the other hand, it has not been possible to find enough PPG (photoplethysmography) data of healthy people and people with arrhythmias. The generation of this data from our watch is not easy either, since we have to count on people with this pathology. Therefore, it is more difficult to create a model that can predict atrial fibrillation, due to lack of data. |
Keywords: | PPG fall detection atrial fibrillation |
Document type: | info:eu-repo/semantics/masterThesis |
Issue Date: | 10-Jan-2021 |
Publication license: | http://creativecommons.org/licenses/by-nc/3.0/es/ |
Appears in Collections: | Bachelor thesis, research projects, etc. |
Files in This Item:
File | Description | Size | Format | |
---|---|---|---|---|
atiendaTFM0121memòria.pdf | Memòria del TFM | 4,15 MB | Adobe PDF | View/Open |
Share:
This item is licensed under a Creative Commons License