This paper proposes cuffless and continuous blood pressure estimation utilising Photoplethysmography (PPG) signals and state of the art recurrent network models, namely, Long Short Term Memory and Gated Recurrent Units. The models were validated on wide range of varying blood pressure and PPG signals acquired from the Multiparameter Intelligent Monitoring in Intensive Care database. Many features were extracted from the PPG waveform and several machine learning techniques were employed in an attempt to eliminate collinearity and reduce the size of input feature vector. Consequently, the most effective features for blood pressure estimation were selected. Experimental results show that the accuracy of the proposed methods outperform traditional models applied in the literature. The results satisfy the American National Standards of the Association for the Advancement of Medical Instrumentation.

Continuous blood pressure monitoring is vital for the early detection and control of hypertension. However, this is unreachable using current non-invasive and cuff-based devices. Several methods have been proposed to overcome this challenge, one of the most attractive ones is the photoplethysmography approach. Although, good performance has been reported in the literature, it was mainly achieved on small inadequate sample size using conventional models that are unable to account for the temporal variations in the input vector. To address these limitations, this paper proposes cuffless and continuous blood pressure estimation using Long Short-term Memory (LSTM) and Gated Recurrent Units (GRU). The models were evaluated on 942 patients acquired from the Multiparameter Intelligent Monitoring in Intensive Care dataset. The proposed models produced superior results in comparison with conventional artificial neural network. In particular, the best performance was achieved by the GRU, with mean absolute error and standard deviation of 5.77±8.52 mmHg and 3.33±5.02 mmHg for systolic and diastolic blood pressure, respectively.