Modeling and Analysis of Tonoarteriogram (TAG) for the Unobtrusive Measurements of Continuous Blood Pressure

Abstract

According to most recent report of the World Health Organization, cardiovascular disease (CVD) has remained the global leading cause of death for the last 20 years and that it is now killing more people than ever before. The steady increases in CVD have become the unprecedented public health crises of international concerns. To fight against the deadly diseases effectively, international communities has proposed a paradigm shift transforming the current reactive medicine to proactive health focusing on the early prediction, early diagnosis and early intervention of diseases. Driven by the great vision of proactive health, mHealth technology enabled by wearable devices has become an emerging field of research and development worldwide.

As the most serious form of CVD, acute cardiovascular events especially ST segment elevation myocardial infarction (STEMI), are life-threatening, time- sensitive emergencies. The overall aim of this thesis was to develop methods suitable for wearable devices that can enable the early prediction of acute CVDs in living-free environments based on models of physiological signals ranging from electrocardiogram (ECG), photoplethysmogram (PPG) to tonoarteriogram (TAG) which is the recording of noninvasive continuous arterial blood pressure.

Emphasis was placed on the importance of focusing on systematic study of TAG in the time and frequency domains and investigation of ultra-short-term (UST) ECG parameters suitable for wearable devices in the diagnosis of STEMI diseases. The ultimate aim was to develop new physiological indexes for CVD prediction and prevention. We set out the following specific objectives to: 1) develop a new mathematical model of TAG signal in the time domain and study the effects of pacing statistics on a newly defined blood pressure variability (BPV); 2) introduce a mathematical spectral model for TAG to facilitate a systematical study on the interactive effects of various factors on the spectrum; 3) investigate the correlation between different modalities of physiological signals; 4) study the extent to which physiological parameters contribute to the spectrum with the emphasis on the development of new frequency indicators of STEMI diseases; 5) exam newly proposed indicators for the performance evaluation of TAG signals/devices; and 6) propose a guideline for the close-loop management of CVDs and COVIDs using wearable-based mobile health (mHealth) technology.

To address the challenges and achieve the specific objectives, we carried out the research activities as follows. First, new mathematical models in both the time and frequency domains for TAG, a continuous arterial blood pressure signal, were developed based on a pacing pulse train (PPT) with random inter-pulse-intervals (IPIs) as the input to the model. The interactive effects of pacing statistics such as heart rate (HR) and heart rate variability (HRV) and individual blood pressure waveform on the newly defined blood pressure variability (BPV) and power density spectrum (PDS) of TAG signal were investigated systematically.

Second, the TAG modeling concepts were extended to theoretical and experimental analysis of ECG. Based on clinical ECG signals from 49 STEMI patients and 42 healthy subjects in PTB Diagnostic Database, we found that besides the interactive effects which are consistent between theoretical and experimental results, the ECG PDSs of STEMI patients exhibited consistently significant power shift towards lower frequency range in ST-elevated leads in comparison with those of reference leads and those of health subjects. The frequency domain alternation of ECG signal in different leads on STEMI patients was evaluated quantitatively using mean frequency (MNF), median frequency (MDF) and their frequency shift ratios. The results of the clinical study showed that the MNF and MDF (p < 0.0001) of ST- elevated leads were significantly lower than those in reference leads by the paired Student’s t-test with the highest median frequency shift ratios at 51.39 ± 12.94% found in anterior MI. To evaluate the accuracy of UST frequency parameters, ECG simulations with systematic changes in PPT firing statistics over various lengths of ECG data ranging from 10s to 60mins were conducted, the results of which revealed that MNF and MDF were less affected by the heart rhythm statistics and the data length but more depended on the alterations of P-QRS-T complexes. These important findings were further confirmed experimentally on 33 more STEMI patients in European ST-T Database, demonstrating that the newly proposed frequency indexes, MNF and MDF as well as their frequency shift ratios, could be potentially used as alternative indicators for STEMI diagnosis even with ultra- short-term ECG recordings of as low as 30s long, so making it suitable for wearable and mHealth applications in living-free environments.

Third, the relationship of consecutive pulse interval series in ECG, PPG and TAG was investigated, and the aforementioned mathematical models were extended to multimodal cardiac signals driven by the same SA node firing which was modeled by a PPT with random IPIs. Models for cross power spectrum and coherence between each two signals were proposed with clinical verification.

Fourth, the performance of various parameters of TAG for continuous BP estimation was investigated in the time domain. Besides the spectral coherence, several new indexes including UST standard deviation of BP signal (SDBP), standard deviation of beat-to-beat diastolic/systolic BPs (SDBBD/SDBBS) of TAG were introduced as scale/parameters to assess the accuracy of continuous BP monitoring devices. Importantly a mathematical model of TAG BPV was derived and verified on the clinical BP data from 26 subjects. Not previously known, the new parameters such as SDBP and RMSD were associated with pulse blood pressure, and they could be independent alternative indexes of CVD and hypertension in addition to systolic and diastolic BPs. The work in this thesis may help to develop the standard for assessing the continuous BP measuring devices, which aims to regulate the potential market of such new devices.

Finally, a wearable-based mHealth workflow for the closed-loop management of acute CVD patients during pandemic was proposed. During a pandemic such as COVID-19, it is critical to balance cardiovascular emergencies with infectious risk. In the context of the proposed mHealth workflow, we recommend using wearable device based mHealth as an early screening and real-time monitoring tool to address this balance and facilitate remote monitoring. The proposed workflow may help to improve the efficiency and effectiveness of a closed-loop management of acute CVD patients especially with STEMI while reducing infection risk.

In summary, this thesis proposed several new models of TAG in both the time and frequency domains, and examined several new physiological parameters including SDBP, RMSD, spectral MNF and MDF and frequency shift ratio which can be easily obtained through wearable devices for potential applications in the diagnosis of acute CVDs especially STEMI in living-free environments. Cross-spectrum analysis was performed between ECG, PPG and TAG signals. Some of newly proposed parameters were investigated together with spectral coherence for the performance evaluation of TAG signals/devices. As an extension, a wearable-based mHealth workflow for the closed-loop management of acute CVD patients during the pandemic was proposed to balance cardiovascular emergencies with infectious risk. Proposed research could help with early prediction and early diagnosis of CVDs and deteriorating health status, saving time for early treatment and reducing both mortality and morbidity.

Date of Award	23 Aug 2021
Original language	English
Awarding Institution	City University of Hong Kong
Supervisor	Yuanting ZHANG (Supervisor)