Tinyml Monitoring Techniques for A-Vent: An Iot Edge for Tracking Clinical Risk Outcomes and Automatic Detection of Patient-Ventilator Asynchrony
CoronaVirus disease 2019 (COVID-19) pandemic, is a respiratory tract infection disease, resulting in high demand for mechanical ventilators. Fronting healthcare workers facing work under pressure and great risk of getting infected due to high demand for care of COVID-19 patients. The COVID-19 pandem...
Saved in:
| Main Author: | |
|---|---|
| Format: | text |
| Published: |
Archīum Ateneo
2021
|
| Subjects: | |
| Online Access: | https://archium.ateneo.edu/theses-dissertations/475 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Institution: | Ateneo De Manila University |
| Summary: | CoronaVirus disease 2019 (COVID-19) pandemic, is a respiratory tract infection disease, resulting in high demand for mechanical ventilators. Fronting healthcare workers facing work under pressure and great risk of getting infected due to high demand for care of COVID-19 patients. The COVID-19 pandemic has prompted open-source developers, engineers, researchers, and institutions to work together to fabricate a low-cost, rapidly deployable mechanical ventilator. As one of the respondents, Ateneo Innovation Center (AIC) of Ateneo de Manila University designed A-Vent, a variation of a low-cost and portable ventilator. As of 2021, the open-source ventilators that have been developed and published lack data telemetry and processing, which are needed for telemedicine capability to support the healthcare system services throughout the current pandemic. This study introduces relevant expertise into embedded systems that provide A-Vent with the real-time patient and ventilator waveform monitoring and telemedicine solutions. Two AI/ML algorithms are used: multilayer perceptron (MLP) is used to recognize early warning score patterns as characteristics and predict clinical risk, a Keras Neural Network is used to interpret the ventilator waveform and detect the emulated patient-ventilator interaction. This was accomplished by the use of embedded machine learning, also known as TinyML, which is the process of deploying the AI/ML models described above on embedded devices. Medical devices using TinyML could interpret biological data in real-time and at a low cost, which could be a solution to the shortage of healthcare personnel as well as a safer way of real-time tracking for patients' clinical information, particularly in the present pandemic. v ACKNOWLEDGEMENTS |
|---|