Unobtrusive home sleep monitoring using wrist-worn wearable photoplethysmography (PPG) could open the way for better sleep disorder screening and health monitoring. However, PPG is rarely included in large sleep studies with gold-standard sleep annotation from polysomnography. Therefore, training data-intensive state-of-the-art deep neural networks is challenging. In this work a deep recurrent neural network is first trained using a large sleep data set with electrocardiogram (ECG) data (292 participants, 584 recordings) to perform 4-class sleep stage classification (wake, rapid-eye-movement, N1/N2, and N3). A small part of its weights is adapted to a smaller, newer PPG data set (60 healthy participants, 101 recordings) through three variations of transfer learning. Best results (Cohen’s kappa of 0.65 ± 0.11, accuracy of 76.36 ± 7.57%) were achieved with the domain and decision combined transfer learning strategy, significantly outperforming the PPG-trained and ECG-trained baselines. This performance for PPG-based 4-class sleep stage classification is unprecedented in literature, bringing home sleep stage monitoring closer to clinical use. The work demonstrates the merit of transfer learning in developing reliable methods for new sensor technologies by reusing similar, older non-wearable data sets. Further study should evaluate our approach in patients with sleep disorders such as insomnia and sleep apnoea.

The electroencephalogram (EEG), for measuring the electrophysiological activity of the brain, has been widely applied in automatic detection of epilepsy seizures. Various EEG-based seizure detection algorithms have already yielded high sensitivity, but training those algorithms requires a large amount of labelled data. Data labelling is often done with a lot of human efforts, which is very time-consuming. In this study, we propose a hybrid system integrating an unsupervised learning (UL) module and a supervised learning (SL) module, where the UL module can significantly reduce the workload of data labelling. For preliminary seizure screening, UL synthesizes amplitude-integrated EEG (aEEG) extraction, isolation forest-based anomaly detection, adaptive segmentation, and silhouette coefficient-based anomaly detection evaluation. The UL module serves to quickly locate the determinate subjects (seizure segments and seizure-free segments) and the indeterminate subjects (potential seizure candidates). Afterwards, more robust seizure detection for the indeterminate subjects is performed by the SL using an EasyEnsemble algorithm. EasyEnsemble, as a class-imbalance learning method, can potentially decrease the generalization error of the seizure-free segments. The proposed method can significantly reduce the workload of data labelling while guaranteeing satisfactory performance. The proposed seizure detection system is evaluated using the Children’s Hospital Boston-Massachusetts Institute of Technology (CHB-MIT) scalp EEG dataset, and it achieves a mean accuracy of 92.62%, a mean sensitivity of 95.55%, and a mean specificity of 92.57%. To the best of our knowledge, this is the first epilepsy seizure detection study employing the integration of both the UL and the SL modules, achieving a competitive performance superior or similar to that of the state-of-the-art methods.

Objective . Evaluate a method for the estimation of the nocturnal systolic blood pressure (SBP) dip from 24 h blood pressure trends using a wrist-worn photoplethysmography (PPG) sensor and a deep neural network in free-living individuals, comparing the deep neural network to traditional machine learning and non-machine learning baselines. Approach . A wrist-worn PPG sensor was worn by 106 healthy individuals for 226 d during which 5111 reference values for blood pressure (BP) were obtained with a 24 h ambulatory BP monitor and matched with the PPG sensor data. Features based on heart rate variability and pulse morphology were extracted from the PPG waveforms. Long- and short term memory (LSTM) networks, dense networks, random forests and linear regression models were trained and evaluated in their capability of tracking trends in BP, as well as the estimation of the SBP dip. Main results . Best performance for estimating the SBP dip were obtained with a deep LSTM neural network with a root mean squared error (RMSE) of 3.12 2.20 mmHg and a correlation of 0.69 . This dip was derived from trend estimates of BP which had an RMSE of 8.22 1.49 mmHg for systolic and 6.55 1.39 mmHg for diastolic BP (DBP). While other models had similar performance for the tracking of relative BP, they did not perform as well as the LSTM for the SBP dip. Significance : The work provides first evidence for the unobtrusive estimation of the nocturnal SBP dip, a highly prognostic clinical parameter. It is also the first to evaluate unobtrusive BP measurement in a large data set of unconstrained 24 h measurements in free-living individuals and provides evidence for the utility of LSTM models in this domain.

Photoplethysmography (PPG) enables unobtrusive heart rate monitoring, which can be used in wrist-worn applications. Its potential for detecting atrial fibrillation (AF) has been recently presented. Besides AF, another cardiac arrhythmia increasing stroke risk and requiring treatment is atrial flutter (AFL). Currently, the knowledge about AFL detection with PPG is limited. The objective of our study was to develop a model that classifies AF, AFL, and sinus rhythm with or without premature beats from PPG and acceleration data measured at the wrist in daily life. A dataset of 40 patients was collected by measuring PPG and accelerometer data, as well as electrocardiogram as a reference, during 24-hour monitoring. The dataset was split into 75%-25% for training and testing a Random Forest (RF) model, which combines features from PPG, inter-pulse intervals (IPI), and accelerometer data, to classify AF, AFL, and other rhythms. The performance was compared to an AF detection algorithm combining traditional IPI features for determining the robustness of the accuracy in presence of AFL. The RF model classified AF/AFL/other with sensitivity and specificity of 97.6/84.5/98.1% and 98.2/99.7/92.8%, respectively. The results with the IPI-based AF classifier showed that the majority of false detections were caused by AFL. The PPG signal contains information to classify AFL in the presence of AF, sinus rhythm, or sinus rhythm with premature contractions. PPG could indicate presence of AFL, not only AF.

Blood pressure (BP) is critical in diagnosing certain cardiovascular diseases such as hypertension. Some previous studies have proved that BP can be estimated by pulse transit time (PTT) calculated by a pair of photoplethysmography (PPG) signals at two body sites. Currently, contact PPG (cPPG) and imaging PPG (iPPG) are two feasible ways to obtain PPG signals. In this study, we proposed a hybrid system (called the ICPPG system) employing both methods that can be implemented on a wearable device, facilitating the measurement of BP in an inconspicuous way. The feasibility of the ICPPG system was validated on a dataset with 29 subjects. It has been proved that the ICPPG system is able to estimate PTT values. Moreover, the PTT measured by the new system shows a correlation on average with BP variations for most subjects, which could facilitate a new generation of BP measurement using wearable and mobile devices.