Waveform Tracker Alarm for Automatic Patient-Ventilator Asynchrony (PVA) and Mechanical State Recognition for Mechanical Ventilators Using Embedded Deep Learning

• Main Authors: Santiago, Paul Ryan A., Cabacungan, Paul M., Oppus, Carlos M, Mamaradlo, John Paul A., Mercado, Neil Angelo M., Cao, Reymond P., Tangonan, Gregory L
• Format: text
• Published: Archīum Ateneo 2024
• Subjects: And circular electronics manufacturing; Biomedical monitoring; Emergency ventilator; Near cloud; TinyML; Biomedical; Computer Engineering; Electrical and Computer Engineering; Engineering
• Online Access: https://archium.ateneo.edu/ecce-faculty-pubs/160; https://doi.org/10.1007/978-981-99-6523-6_10
• Institution: Ateneo De Manila University

The Ateneo Innovation Center designs and develops a modular approach to medical alarm and alert systems for mechanical ventilators that enable clinicians to remotely monitor patient conditions and ventilator circuit status in near real-time, providing decision support that allows for a better diagnosis. It monitors and tracks the alarm events related to the ventilator waveform consisting of pressure, flow, and volume curves by using automatic peak detection of the curves and real-time recognition of time-series waveforms. The developed system combines the threshold alarms with embedded Artificial Intelligence to automatically detect complex alarms that need medical expertise such as issue detection on asynchrony, anomalies, and mechanical. It also differentiates the critical types of alarms, assisting clinicians via alarm prioritization, and remote patient monitoring via a near cloud system. Storing data in the near cloud system as a medical database enables building a rich dataset for upgrading the predictive model of alarm recognition.