A knowledge-based approach to arterial stiffness estimation using the digital volume pulse

Predicting Arterial Stiffness From Single-Channel Photoplethysmography Signal: A Feature Interaction-Based Approach

Arterial stiffness (AS) serves as a crucial indicator of arterial elasticity and function, typically requiring expensive equipment for detection. Given the strong correlation between AS and various photoplethysmography (PPG) features, PPG emerges as a convenient method for assessing AS. However, the limitations of independent PPG features hinder detection accuracy. This study introduces a feature selection method leveraging the interactive relationships between features to enhance the accuracy of predicting AS from a single-channel PPG signal. Initially, an adaptive signal interception method was employed to capture high-quality signal fragments from PPG sequences. 58 PPG features, deemed to have potential contributions to AS estimation, were extracted and analyzed. Subsequently, the interaction factor (IF) was introduced to redefine the interaction and redundancy between features. A feature selection algorithm (IFFS) based on the IF was then proposed, resulting in a combination of interactive features. Finally, the Xgboost model is utilized to estimate AS from the selected features set. The proposed approach is evaluated on datasets of 268 male and 124 female subjects, respectively. The results of AS estimation indicate that IFFS yields interacting features from numerous sources, rejects redundant ones, and enhances the association. The interaction features combined with the Xgboost model resulted in an MAE of 122.42 and 142.12 cm/sec, an SDE of 88.16 and 102.56 cm/sec, and a PCC of 0.88 and 0.85 for the male and female groups, respectively. The findings of this study suggest that the stated method improves the accuracy of predicting AS from single-channel PPG, which can be used as a non-invasive and cost-effective screening tool for atherosclerosis.

Estimation of aortic stiffness by finger photoplethysmography using enhanced pulse wave analysis and machine learning

Introduction

Aortic stiffness plays a critical role in the evolution of cardiovascular diseases, but the assessment requires specialized equipment. Photoplethysmography (PPG) and single-lead electrocardiogram (ECG) are readily available in healthcare and wearable devices. We studied whether a brief PPG registration, alone or in combination with single-lead ECG, could be used to reliably estimate aortic stiffness.

Methods

A proof-of-concept study with simultaneous high-resolution index finger recordings of infrared PPG, single-lead ECG, and finger blood pressure (Finapres) was performed in 33 participants [median age 44 (range 21-66) years, 19 men] and repeated within 2 weeks. Carotid-femoral pulse wave velocity (cfPWV; two-site tonometry with SphygmoCor) was used as a reference. A brachial single-cuff oscillometric device assessed aortic pulse wave velocity (aoPWV; Arteriograph) for further comparisons. We extracted 136 established PPG waveform features and engineered 13 new with improved coupling to the finger blood pressure curve. Height-normalized pulse arrival time (NPAT) was derived using ECG. Machine learning methods were used to develop prediction models.

Results

The best PPG-based models predicted cfPWV and aoPWV well (root-mean-square errors of 0.70 and 0.52 m/s, respectively), with minor improvements by adding NPAT. Repeatability and agreement were on par with the reference equipment. A new PPG feature, an amplitude ratio from the early phase of the waveform, was most important in modelling, showing strong correlations with cfPWV and aoPWV (r = -0.81 and -0.75, respectively, both P < 0.001).

Conclusion

Using new features and machine learning methods, a brief finger PPG registration can estimate aortic stiffness without requiring additional information on age, anthropometry, or blood pressure. Repeatability and agreement were comparable to those obtained using non-invasive reference equipment. Provided further validation, this readily available simple method could improve cardiovascular risk evaluation, treatment, and prognosis.

An XGBoost-based model for assessment of aortic stiffness from wrist photoplethysmogram

BACKGROUND AND OBJECTIVE

Carotid-femoral pulse wave velocity (cf-PWV) is the gold standard for non-invasive assessment of aortic stiffness. Photoplethysmography used in wearable devices provides an indirect measurement method for cf-PWV. This study aimed to construct a cf-PWV prediction method based on the XGBoost algorithm and wrist photoplethysmogram (wPPG) for the early screening of arteriosclerosis in primary healthcare.

METHODS

Data from 210 subjects were used for modeling, and 100 subjects were used as an external validation set. The wPPG pulse waves were filtered by discrete wavelet transform, and various features were extracted from each waveform, including two original indexes. The extraction rate (ER) and Pearson P were calculated to evaluate the applicability of each feature for model training. The magnitude of cf-PWV was predicted by an XGBoost-based model using the selected features and basic physiological parameters (age, sex, height, weight and BMI). The level of aortic stiffness was classified by a 3-classification strategy according to the standard cf-PWV (measured by the Complior device). Bland-Altman plot, Pearson correlation analysis, and accuracy tested performance from two aspects: predicting the magnitude of cf-PWV and classifying the level of aortic stiffness.

RESULTS

In the external validation set (n = 100, age range 22-79), 97 subjects obtained features (ER = 97%). The predicted cf-PWV was significantly correlated with the standard cf-PWV (r = 0.927, P < 0.001). The accuracy (AC) of the 3-classification was 85.6%. The interrater agreement for assessing aortic stiffness was at least substantial (quadratically weighted Kappa = 0.833).

CONCLUSIONS

The multi-parameter fusion cf-PWV prediction method based on the XGBoost algorithm and wPPG pulse wave analysis proves the feasibility of atherosclerosis screening in wearable devices.

Intelligent Oscillometric System for Automatic Detection of Peripheral Arterial Disease

Peripheral arterial disease (PAD) is a progressing arterial disorder that is associated with significant morbidity and mortality. The conventional PAD detection methods are invasive, cumbersome, or require expensive equipment and highly trained technicians. Here, we propose a new automated, noninvasive, and easy-to-use method for the detection of PAD based on characterizing the arterial system by applying an external varying pressure using a cuff. The superposition of the internal arterial pressure and the externally applied pressure were measured and mathematically modeled as a function of cuff pressure. A feature-based learning algorithm was then designed to identify PAD patterns by analyzing the parameters of the derived mathematical models. Genetic algorithm and principal component analysis were employed to select the best predictive features distinguishing PAD patterns from normal. A RUSBoost ensemble model using neural network as the base learner was designed to diagnose PAD from genetic algorithm selected features. The proposed method was validated on data collected from 14 PAD patients and 19 healthy individuals. It achieved a high accuracy, sensitivity, and specificity of 91.4%, 90.0%, and 92.1%, respectively, in detecting PAD. The effect of age, a confounding factor that may have impacted our analyzes, was not considered in this study. The proposed method shows promise toward noninvasive and accurate detection of PAD and can be integrated into routine oscillometric blood pressure measurements.

Smart automated heart health monitoring using photoplethysmography signal classification

This paper proposes a smart, automated heart health-monitoring (SAHM) device using a single photoplethysmography (PPG) sensor that can monitor cardiac health. The SAHM uses an Orthogonal Matching Pursuit (OMP)-based classifier along with low-rank motion artifact removal as a pre-processing stage. Major contributions of the proposed SAHM device over existing state-of-the-art technologies include these factors: (i) the detection algorithm works with robust features extracted from a single PPG sensor; (ii) the motion compensation algorithm for the PPG signal can make the device wearable; and (iii) the real-time analysis of PPG input and sharing through the Internet. The proposed low-cost, compact and user-friendly PPG device can also be prototyped easily. The SAHM system was tested on three different datasets, and detailed performance analysis was carried out to show and prove the efficiency of the proposed algorithm.

A Simple and Robust Method for Determining the Quality of Cardiovascular Signals Using the Signal Similarity

This paper proposes a novel signal quality assessment method for quasi-periodic cardiovascular signals, chiefly focus on the photoplethysmogram (PPG). The proposed method utilizes the fact that most cardiovascular signals are slowly time varying and thus morphological aspects of the two adjacent beats are almost identical. In order to implement this idea, the method first identifies pulse onset to divide the signal into several segments each of which contains one period of the signal. The segmented pulse signals having different pulse durations are then temporarily normalized by resampling them at a specific rate. Finally, the quality of the signals is evaluated as the signal similarity between the two adjacent segments. Optimal thresholds for the classification between high-and low-quality PPG signals are determined using the equal training sensitivity and specificity criterion. The proposed method is evaluated using a database where PPG signals are collected during a variety of activities such as cycling exercise. It attains a sensitivity of 97.9%, a specificity of 85.3%, and an accuracy of 93.8%, compared to manually annotated results. The promising results indicate that the proposed method is affordable to simply determine the quality of quasi-periodic cardiovascular signals, particularly PPG signals. In addition, based on the quasi-periodic characteristics of cardiovascular signals, the proposed method can also be used to indicate the reliability and the availability of the collected signals.

Use of zero-frequency resonator for automatically detecting systolic peaks of photoplethysmogram signal

This work investigates the application of zero-frequency resonator (ZFR) for detecting systolic peaks of photoplethysmogram (PPG) signals. Based on the authors' studies, they propose an automated noise-robust method, which consists of the central difference operation, the ZFR, the mean subtraction and averaging, the peak determination, and the peak rejection/acceptance rule. The method is evaluated using different kinds of PPG signals taken from the standard MIT-BIH polysomnographic database and Complex Systems Laboratory database and the recorded PPG signals at their Biomedical System Lab. The method achieves an average sensitivity (Se) of 99.95%, positive predictivity (Pp) of 99.89%, and overall accuracy (OA) of 99.84% on a total number of 116,673 true peaks. Evaluation results further demonstrate the robustness of the ZFR-based method for noisy PPG signals with a signal-to-noise ratio (SNR) ranging from 30 to 5 dB. The method achieves an average Se = 99.76%, Pp = 99.84%, and OA = 99.60% for noisy PPG signals with a SNR of 5 dB. Various results show that the method yields better detection rates for both noise-free and noisy PPG signals. The method is simple and reliable as compared with the complexity of signal processing techniques and detection performance of the existing detection methods.

PPG pulse direction determination algorithm for PPG waveform inversion by wrist rotation

This paper describes photoplethysmography (PPG)-based pulse direction determination algorithm on a site of the radial artery using a wrist band. It has been well known that PPG is susceptible to noise and motion artifacts in the mobile environment and many research efforts have been made to focus on rejection of the noise and motion artifacts. However, no research has been performed to find PPG pulses when PPG is inverted by wrist movement. We present an algorithm, which accurately yields which direction PPG pulses face regardless of wrist movement. The algorithm is one step closer to robust real-time PPG pulse direction determination for continuous PPG monitoring regardless of body movements.

A modified D-max method to estimate heart rate at a ventilatory threshold during an incremental exercise test

The purpose of this study was to design a modified D-max method to determine heart rate at a ventilatory threshold (HRVT) and to investigate whether this method would be valid during incremental exercise tests. The HRVT was estimated from a new parameter defined as HR at the maximal difference point between linearly- and quadratically approximated HR trends (modified D-max method). HR and ventilatory gas data for 105 subjects (53 males and 52 females; 38.26 ± 12.06 years; 166.62 ± 8.21 cm; 65.31 ± 11.10 kg) were simultaneously collected during an incremental treadmill test to evaluate the validity of the modified D-max method. Reference HRVTs were manually identified from the ventilatory gas data by an experienced sports physiologist and compared with those estimated by the HR parameter. A strong positive correlation (r = 0.71, p <; 0.01) and a low HR difference of 9.94 ± 7.10 bpm between the reference and estimated HRVTs were obtained. The results indicate that the modified D-max method outperforms the conventional D-max method (r = 0.53, p <; 0.01), the three-piece linear regression lines method (r = 0.42, p <; 0.01), and the parallel straight line slope method (r = 0.57, p <; 0.01). Furthermore, the modified D-max method improves the predictive accuracy of HRVTs by combining its result with subject's age. The combined parameters have a strong positive correlation with the reference HRVTs (r = 0.74, p <; 0.01) and a lower HR difference of 9.40 ± 6.91 bpm. The results suggest that the modified D-max method is highly applicable to predicting HRVTs during incremental exercise tests and also improves HRVT detection accuracy by combining its result with the subject's age.

A Novel AMARS Technique for Baseline Wander Removal Applied to Photoplethysmogram

A new digital filter, AMARS (aligning minima of alternating random signal) has been derived using trigonometry to regulate signal pulsations inline. The pulses are randomly presented in continuous signals comprising frequency band lower than the signal's mean rate. Frequency selective filters are conventionally employed to reject frequencies undesired by specific applications. However, these conventional filters only reduce the effects of the rejected range producing a signal superimposed by some baseline wander (BW). In this work, filters of different ranges and techniques were independently configured to preprocess a photoplethysmogram, an optical biosignal of blood volume dynamics, producing wave shapes with several BWs. The AMARS application effectively removed the encountered BWs to assemble similarly aligned trends. The removal implementation was found repeatable in both ear and finger photoplethysmograms, emphasizing the importance of BW removal in biosignal processing in retaining its structural, functional and physiological properties. We also believe that AMARS may be relevant to other biological and continuous signals modulated by similar types of baseline volatility.

A preliminary study of a running speed based heart rate prediction during an incremental treadmill exercise

This preliminary study investigates feasibility of a running speed based heart rate (HR) prediction. It is basically motivated from the assumption that there is a significant relationship between HR and the running speed. In order to verify the assumption, HR and running speed data from 217 subjects of varying aerobic capabilities were simultaneously collected during an incremental treadmill exercise. A running speed was defined as a treadmill speed and its corresponding heart rate was calculated by averaging the last one minute HR values of each session. The feasibility was investigated by assessing a correlation between the heart rate and the running speed using inter-subject (between-subject) and intra-subject (within-subject) datasets with regression orders of 1, 2, 3, and 4, respectively. Furthermore, HR differences between actual and predicted HRs were also employed to investigate the feasibility of the running speed in predicting heart rate. In the inter-subject analysis, a strong positive correlation and a reasonable HR difference (r = 0.866, 16.55±11.24 bpm @ 1st order; r = 0.871, 15.93±11.49 bpm @ 2nd order; r = 0.897, 13.98±10.80 bpm @ 3rd order; and r = 0.899, 13.93±10.64 bpm @ 4th order) were obtained, and a very high positive correlation and a very low HR difference (r = 0.978, 6.46±3.89 bpm @ 1st order; r = 0.987, 5.14±2.87 bpm @ 2nd order; r = 0.996, 2.61±2.03 bpm @ 3rd order; and r = 0.997, 2.04±1.73 bpm @ 4th order) were obtained in the intra-subject analysis. It can therefore be concluded that 1) heart rate is highly correlated with a running speed; 2) heart rate can be approximately estimated by a running speed with a proper statistical model (e.g., 3rd-order regression); and 3) an individual HR-speed calibration process may improve the prediction accuracy.

A robust method for pulse peak determination in a digital volume pulse waveform with a wandering baseline

This paper presents a robust method for pulse peak determination in a digital volume pulse (DVP) waveform with a wandering baseline. A proposed new method uses a modified morphological filter (MMF) to eliminate a wandering baseline signal of the DVP signal with minimum distortion and a slope sum function (SSF) with an adaptive thresholding scheme to detect pulse peaks from the baseline-removed DVP signal. Further in order to cope with over-detected and missed pulse peaks, knowledge based rules are applied as a postprocessor. The algorithm automatically adjusts detection parameters periodically to adapt to varying beat morphologies and fluctuations. Compared with conventional methods (highpass filtering, linear interpolation, cubic spline interpolation, and wavelet adaptive filtering), our method performs better in terms of the signal-to-error ratio, the computational burden (0.125 seconds for one minute of DVP signal analysis with the Intel Core 2 Quad processor @ 2.40 GHz PC), the true detection rate (97.32% with an acceptance level of 4 ms ) as well as the normalized error rate (0.18%). In addition, the proposed method can detect true positions of pulse peaks more accurately and becomes very useful for pulse transit time (PTT) and pulse rate variability (PRV) analyses.

Shape-preserving preprocessing for human pulse signals based on adaptive parameter determination

The use of the human pulse signal for medical diagnosis is a mainstay in the practice of traditional Chinese medicine. Computer processing of this signal may be used to automate diagnostic procedures and to reveal sources of information in the waveform that have been used by both eastern and western physicians for more than two millennia. A new method for preprocessing of the human pulse signal significantly improves feature extraction and classification of the waveform. Baseline distortion is first removed using the dual-tree complex wavelet transform (DT-CWT) and cubic spline interpolation, then a novel filtering method removes the residual background noise. Filtering is implemented in two stages. In the initial pass, a majority of the noise is eliminated by an adaptive mean filter whose sliding window duration is selected automatically based on a chain code and the DT-CWT. In the second pass, residual high frequency noise is removed using the DT-CWT with a new threshold determination. Experimental results demonstrate effective removal of background disturbances with excellent preservation of pulse peak information essential for proper parametric representation and classification of the waveform.