Nombre: LÍVIA GONÇALVES GOMES
Tipo: MSc dissertation
Fecha de publicación: 26/10/2023
Supervisor:
Nombre |
Papel |
|---|---|
| ARNALDO GOMES LEAL JÚNIOR | Advisor * |
| RICARDO CARMINATI DE MELLO | Co-advisor * |
Junta de examinadores:
| Nombre |
Papel |
|---|---|
| ARNALDO GOMES LEAL JÚNIOR | Advisor * |
| CAMILO ARTURO RODRIGUEZ DIAZ | Internal Examiner * |
| CARLOS ALBERTO FERREIRA MARQUES | External Examiner * |
| RICARDO CARMINATI DE MELLO | Co advisor * |
Sumario: Respiratory rate is one of the important physiological signals used to monitor human health in disease and physical activity areas. In this context, there are multiparameter sensors that are commercially known and usually applied in hospital settings. Their design does not provide great mobility for the patient, in addition to the high cost, making accessibility to other environments
difficult. On the other hand, the advancement of technologies has led to the development of various types of electrical sensors for measuring these signals; however, the majority of them are electrically and electronically based, not being suitable for environments with electromagnetic interference. This work presents a respiratory rate system composed of a polymer optical fiber
sensor using a smartphone as the interrogator system. Additionally, integration with the Internet of Things was carried out, using Edge Computing techniques for local signal processing in an application developed on the AndroidStudio platform. The ThingSpeak platform is used for cloud storage and the ThingView mobile application allows online viewing of information, providing the remote access. To verify the accuracy of the sensor, a metronome was set to specific frequency rates, used as a reference. The sensor underwent extension and retraction movements, carried out manually, at the frequency of the metronomes beats at these rates. The sensory system presents a maximum percentage error of 4.5% (1.35 BPM - Breaths per Minute) when comparing
the values obtained with the reference frequencies used. Furthermore, tests were performed on volunteers to verify the systems performance in a real environment, WHERE they were asked to breathe normally, at rest, and also simulating the practice of physical exercises at three known breathing rates. The highest percentage error verified for the resting state is 3.63% (0.8 BPM)
and for the moving state 5.35% (1.88 BPM). To verify the remote access to information, the visualization of respiratory rate data measured by the local system in ThingView was analyzed. The frequency read in the application is the same measured by the sensor, presented instantly, it is also visualized the measurement information of the week and month, showing the efficiency of
the proposed approach for remote sensing applications with cloud integration.
