Acoustic sensing as a novel approach for cardiovascular monitoring at the wrist

Abstract

Cardiovascular diseases are the number one cause of deaths globally. An increased cardiovascular risk can be detected by a regular monitoring of the vital signs including the heart rate, the heart rate variability (HRV) and the blood pressure. For a user to undergo continuous vital sign monitoring, wearable systems prove to be very useful as the device can be integrated into the user's lifestyle without affecting the daily activities. However, the main challenge associated with the monitoring of these cardiovascular parameters is the requirement of different sensing mechanisms at different measurement sites. There is not a single wearable device that can provide sufficient physiological information to track the vital signs from a single site on the body. This thesis proposes a novel concept of using acoustic sensing over the radial artery to extract cardiac parameters for vital sign monitoring. A wearable system consisting of a microphone is designed to allow the detection of the heart sounds together with the pulse wave, an attribute not possible with existing wrist-based sensing methods.

Methods: The acoustic signals recorded from the radial artery are a continuous reflection of the instantaneous cardiac activity. These signals are studied and characterised using different algorithms to extract cardiovascular parameters. The validity of the proposed principle is firstly demonstrated using a novel algorithm to extract the heart rate from these signals. The algorithm utilises the power spectral analysis of the acoustic pulse signal to detect the S1 sounds and additionally, the K-means method to remove motion artifacts for an accurate heartbeat detection. The HRV in the short-term acoustic recordings is found by extracting the S1 events using the relative information between the short- and long-term energies of the signal. The S1 events are localised using three different characteristic points and the best representation is found by comparing the instantaneous heart rate profiles. The possibility of measuring the blood pressure using the wearable device is shown by recording the acoustic signal under the influence of external pressure applied on the arterial branch. The temporal and spectral characteristics of the acoustic signal are utilised to extract the feature signals and obtain a relationship with the systolic blood pressure (SBP) and diastolic blood pressure (DBP) respectively.

Results: This thesis proposes three different algorithms to find the heart rate, the HRV and the SBP/ DBP readings from the acoustic signals recorded at the wrist. The results obtained by each algorithm are as follows: 1. The heart rate algorithm is validated on a dataset consisting of 12 subjects with a data length of 6 hours. The results demonstrate an accuracy of 98.78%, mean absolute error of 0.28 bpm, limits of agreement between -1.68 and 1.69 bpm, and a correlation coefficient of 0.998 with reference to a state-of-the-art PPG-based commercial device. A high statistical agreement between the heart rate obtained from the acoustic signal and the photoplethysmography (PPG) signal is observed. 2. The HRV algorithm is validated on the short-term acoustic signals of 5-minutes duration recorded from each of the 12 subjects. A comparison is established with the simultaneously recorded electrocardiography (ECG) and PPG signals respectively. The instantaneous heart rate for all the subjects combined together achieves an accuracy of 98.50% and 98.96% with respect to the ECG and PPG signals respectively. The results for the time-domain and frequency-domain HRV parameters also demonstrate high statistical agreement with the ECG and PPG signals respectively. 3. The algorithm proposed for the SBP/ DBP determination is validated on 104 acoustic signals recorded from 40 adult subjects. The experimental outputs when compared with the reference arm- and wrist-based monitors produce a mean error of less than 2 mmHg and a standard deviation of error around 6 mmHg.

Based on these results, this thesis shows the potential of this new sensing modality to be used as an alternative, or to complement existing methods, for the continuous monitoring of heart rate and HRV, and spot measurement of the blood pressure at the wrist.