Estimation of blood pressure using Heart Rate Variability features with deep learning approach for limited database

Blood pressure (BP) measurement gains particular relevance in regular fitness activities and is important for the detection of cardiovascular diseases. Cardiovascular disease is a term that encompasses the serious risks of cardiopathy, stroke, renal failure, and other health issues. When blood flow changes the pressure on blood arteries changes which cause hypertension(high blood pressure) or hypotension(low blood pressure). . Hypertension and Hypotension are dangerous illnesses since they can damage multiple organs if left untreated for a long time. Such a life-threatening situation necessitates rapid clinical intervention.Normally, blood pressure is measured using a cuff on the arm, and this method is extensively utilized in practically every hospital. However, for continuous measurement, the standard inflated cuff creates pain and clumsiness in the patient, which might result in a misleading reading of blood pressure. As a result, the researchers favored BP measuring methods that did not require a cuff. Continuous blood pressure monitoring in ICU patients can greatly aid in determining the risk of developing hypotension or hypertension. Many studies have extracted characteristics from numerous physiological signals such as ECG, PPG, Impedance Cardiography(ICG), and IPG to assess BP. However, training extraneous features taken from these physiological signals has an impact on model performance and increases model complexity. Existing cuff-less BP measuring research yields either normal or hypertension readings, resulting in distinct model creation for different BP scenarios. To circumvent these constraints, a novel learning-based methodology for continuous BP monitoring has been presented.