Continuous vital signals monitoring has gained a huge relevance for disease prevention that afflict a large part of the world population, for this reason, the healthcare equipment should be easy-wear and convenient-operate. Nonintrusive and noninvasive detective methods are the basic requirement for the wearable medical devices, especially when the devices are used in sports applications or by the elderly for self-monitoring. The aim of this paper is to measure continuous arterial blood pressure (ABP) through a cuffless non-intrusive approach. The arterial blood pressure is an essential physiological parameter for health monitoring. Most blood pressure measurement devices determine the systolic and diastolic arterial blood pressure through the inflation and the deflation of a cuff. This method is uncomfortable to the user and may cause anxiety which in turns can affect the blood pressure. The approach utilized in this paper is based on deep learning techniques: different neural networks are used to infer ABP starting from photoplethysmogram (PPG) and electrocardiogram (ECG). In particular, ABP was predicted first utilizing only PPG and then both PPG and ECG. It was demonstrated that adding ECG improved performance in every configuration achieving, after personalization, a MAE equal to 4.118 mmHg on systolic blood pressure and 2.228 mmHg on diastolic blood pressure with a ResNet followed by 3 LSTM layers. Results are compliant with the American National Standards of the Association for the Advancement of Medical Instrumentation. ECG, PPG and ABP measurements are extracted from the MIMIC database that contains clinical signal data reflecting real measurements and validates the results on a custom dataset created at Neuronica Lab, Politecnico di Torino.
AUTHOR: Stefano Villata
ADVISORS: Eros Gian Alessandro Pasero, Annunziata Paviglianiti
DEGREE COURSE: Master’s Degree in Biomedical Engineering
ACADEMIC YEAR: 2020/2021