The Deep-Match Framework: R-Peak Detection in Ear-ECG.

The Ear-ECG provides a continuous Lead I like electrocardiogram (ECG) by measuring the potential difference related to heart activity by electrodes which are embedded within earphones. However, the significant increase in wearability and comfort enabled by Ear-ECG is often accompanied by a degradation in signal quality - an obstacle that is shared by the majority of wearable technologies. We aim to resolve this issue by introducing a Deep Matched Filter (Deep-MF) for the highly accurate detection of R-peaks in wearable ECG, thus enhancing the utility of Ear-ECG in real-world scenarios. The Deep-MF consists of an encoder stage, partially initialised with an ECG template, and an R-peak classifier stage. Through its operation as a Matched Filter, the encoder searches for matches with an ECG template in the input signal, prior to filtering these matches with the subsequent convolutional layers and selecting peaks corresponding to the ground-truth ECG. The latent representation of R-peak information is then fed into a R-peak classifier, of which the output provides precise R-peak locations. The proposed Deep Matched Filter is evaluated using leave-one-subject-out cross-validation over 36 subjects with an age range of 18-75, with the Deep-MF outperforming existing algorithms for R-peak detection in noisy ECG. The Deep-MF achieves a median R-peak recall of 94.9% and a median precision of 91.2% across subjects when evaluated with leave-one-subject-out cross validation. Overall, this Deep-Match framework serves as a valuable step forward for the real-world functionality of Ear-ECG and, through its interpretable operation, the acceptance of deep learning models in e-Health.

Generalization Error of a Regression Model for Non-Invasive Blood Pressure Monitoring using a Single Photoplethysmography (PPG) Signal.

Photoplethysmography (PPG) sensors integrated in wearable devices offer the potential to monitor arterial blood pressure (ABP) in patients. Such cuffless, non-invasive, and continuous solution is suitable for remote and ambulatory monitoring. A machine learning model based on PPG signal can be used to detect hypertension, estimate beat-by-beat ABP values, and even reconstruct the shape of the ABP. Overall, models presented in literature have shown good performance, but there is a gap between research and potential real-world use cases. Usually, models are trained and tested on data from the same dataset and same subjects, which may lead to overestimating their accuracy. In this paper: we compare cross-validation, where the test data are from the same dataset as training data, and external validation, where the model is tested on samples from a new dataset, on a regression model which predicts diastolic blood pressure from PPG features. The results show that, in the cross-validation, the predicted and the real values are linearly dependent, while in the external validation, the predicted values are not related to the real ones, but probably just through an average value.

Feasibility of Transfer Learning from Finger PPG to In-Ear PPG.

The success of deep learning methods has enabled many modern wearable health applications, but has also highlighted the critical caveat of their extremely data hungry nature. While the widely explored wrist and finger photoplethysmography (PPG) sites are less affected, given the large available databases, this issue is prohibitive to exploring the full potential of novel recording locations such as in-ear wearables. To this end, we assess the feasibility of transfer learning from finger PPG to in-ear PPG in the context of deep learning for respiratory monitoring. This is achieved by introducing an encoder-decoder framework which is set up to extract respiratory waveforms from PPG, whereby simultaneously recorded gold standard respiratory waveforms (capnography, impedance pneumography and air flow) are used as a training reference. Next, the data augmentation and training pipeline is examined for both training on finger PPG and the subsequent fine tuning on in-ear PPG. The results indicate that, through training on two large finger PPG data sets (95 subjects) and then retraining on our own small in-ear PPG data set (6 subjects), the model achieves lower and more consistent test error for the prediction of the respiratory waveforms, compared to training on the small in-ear data set alone. This conclusively demonstrates the feasibility of transfer learning from finger PPG to in-ear PPG, leading to better generalisation across a wide range of respiratory rates.

Design and Implementation of an Atrial Fibrillation Detection Algorithm on the ARM Cortex-M4 Microcontroller.

At present, a medium-level microcontroller is capable of performing edge computing and can handle the computation of neural network kernel functions. This makes it possible to implement a complete end-to-end solution incorporating signal acquisition, digital signal processing, and machine learning for the classification of cardiac arrhythmias on a small wearable device. In this work, we describe the design and implementation of several classifiers for atrial fibrillation detection on a general-purpose ARM Cortex-M4 microcontroller. We used the CMSIS-DSP library, which supports Naïve Bayes and Support Vector Machine classifiers, with different kernel functions. We also developed Python scripts to automatically transfer the Python model (trained in Scikit-learn) to the C environment. To train and evaluate the models, we used part of the data from the PhysioNet/Computing in Cardiology Challenge 2020 and performed simple classification of atrial fibrillation based on heart-rate irregularity. The performance of the classifiers was tested on a general-purpose ARM Cortex-M4 microcontroller (STM32WB55RG). Our study reveals that among the tested classifiers, the SVM classifier with RBF kernel function achieves the highest accuracy of 96.9%, sensitivity of 98.4%, and specificity of 95.8%. The execution time of this classifier was 720 µs per recording. We also discuss the advantages of moving computing tasks to edge devices, including increased power efficiency of the system, improved patient data privacy and security, and reduced overall system operation costs. In addition, we highlight a problem with false-positive detection and unclear significance of device-detected atrial fibrillation.

Exaggerated pressor response to static squats in Parkinson's disease (PD) and healthy subjects is likely an individual trait, not influenced by whole body vibration (WBV).

BACKGROUND: Reduced muscle strength is one symptom of Parkinson's disease (PD). Strength can be increased by strength training, which may cause exaggerated blood pressure (BP) rise. It is believed that exercises performed on vibrating platform can strengthen leg muscles without excessive BP increase. OBJECTIVE: To measure the pressor response to static exercises performed during whole body vibration in PD patients. METHODS: Twenty-four aged PD patients and twelve healthy young volunteers participated in the study. PD subjects performed six repetitions of deep-, semi-squat, and calves at vibration frequency of 30 Hz. Each 30 s exercise was followed by 30 s rest. The young volunteers performed two sessions of above-mentioned exercises with and without vibration. BP was measured continuously. RESULTS: In PD patients, the highest BP values were observed during deep squat; systolic blood pressure rose 10 mmHg in 'weak responders', and 50 mmHg in 'strong responders'. This difference correlated with the rise in pulse pressure suggesting indirectly the role of stoke volume in individual response. In healthy subjects pressor response was also individually differentiated and not influenced by vibration. CONCLUSION: Deep and semi squat can evoke a strong cardiovascular response in some PD and healthy subjects. Low-magnitude vibrations likely did not affect pressor response.

Device for Controlling Stimulus Self-Application During Autonomic Nervous System Tests.

Assessment of autonomic nervous system (ANS) functioning may be performed non-invasively using autonomic tests which are based on evaluation of response of cardiovascular system to the applied stimuli, such as increased air pressure during Valsalva maneuver, skeletal muscle contraction during static handgrip or deep slow breathing. The cardiovascular response depends, besides ANS reaction and test protocol, also on the way stimulus is self-applied by the test subject. We present a versatile device for controlling stimulus self-application during three ANS tests: Valsalva maneuver, static handgrip, and deep breathing. It integrates two different gauges and a pace setter for breathing into one device. The core of the device is a linear LED display which, using green, yellow, and red diodes, informs the subject about the correctness of self-application of respective stimulus. The settings of the device can be adjusted to the needs of the protocol chosen. The device can record the duration of mouth air pressure or the force produced by the subject during ANS tests, which assures correctness of the tests, thus allowing to track individual variability changes in the response to the test. The device was verified during ANS tests and its use was intuitive for patients, reducing the time needed for training before tests and decreasing the effort of the physician.

Assessment of calibration methods on impedance pneumography accuracy.

The aim was to assess accuracy of tidal volumes (TV) calculated by impedance pneumography (IP), reproducibility of calibration coefficients (CC) between IP and pneumotachometry (PNT), and their relationship with body posture, breathing rate and depth. Fourteen students performed three sessions of 18 series: normal and deep breathing at 6, 10, 15 breaths/min rates, while supine, sitting and standing; 18 CC were calculated for every session. Session 2 was performed 2 months after session 1, session 3 1-3 days after session 2. TV were calculated using full or limited set of CC from current session, in case of sessions 2 and 3 also using CC from session 1 and 2, respectively. When using full set of CC from current session, IP underestimated TV by -3.2%. Using CC from session 2 for session 3 measurements caused decrease of relative difference: -3.9%, from session 1 for session 2: -5.3%; for limited set of CC: -5.0%. The body posture had significant effect on CC. The highest accuracy was obtained when all factors influencing CC were considered. The application of CC related only to body posture may result in shortening of calibration and moderate accuracy loss. Using CC from previous session compromises accuracy moderately.