Physiological measurement最新文献

Objective.Wearable devices are becoming increasingly prevalent, offering the capability to estimate energy expenditure. Among these devices, the Apple Watch has demonstrated notable results in estimating energy expenditure during physical activity, especially compared to other wearable devices. Its accuracy in determining energy expenditure during electromyostimulation (EMS) training remains unexplored and is investigated in this work.Methods.35 young, healthy adults completed two stepwise increasing bike ergometer protocols (50/30/3 protocol) until the maximum physical load was reached with and without EMS support. Energy expenditure estimates from the Apple Watch Series 7 (Apple Inc., Cupertino, California, USA) were compared against gold-standard spirometric calorimetry measurements.Results.The Apple Watch Series 7 underestimated energy expenditure compared to spirometric calorimetry for all data (mean difference: -27.4 kcal, LoA: 62.2 kcal), for ergometer exercise without EMS (mean difference: -28.8 kcal, LoA: 62.8 kcal), and for ergometer exercise with EMS (mean difference: -26.0 kcal, LoA: 62.4 kcal) data. We observed strong correlations between the Apple Watch Series 7 and spirometric calorimetry, withr= 0.93 (p< 0.001) for all data,r= 0.93 (p< 0.001) for exercise without EMS, andr= 0.93 (p< 0.001) for exercise with EMS.Conclusion.The Apple Watch Series 7 showed consistent accuracy in estimating energy expenditure during ergometer exercise, both with and without EMS. These findings suggest that the device can reliably monitor energy expenditure during EMS training, exhibiting similar accuracy limitations to conventional exercise settings.

Objective. The aim of this study was to explore free-breathing cardiac cine images reconstructed with sensor-free physiological signals estimates. Such signals were estimated using the noise variance of the radio frequency receiver coils. Reconstructions with reference signals acquired during MR scan were compared with the sensor-free reconstructions using an extended CineJENSE algorithm.Approach. Free-breathing untriggered MRI cine data from 27 patients and 22 healthy volunteers in various slice orientations were acquired simultaneously with physiological signals using external sensors (ECG and respiratory belts). Physiological signals were estimated using the noise variance of receiver coils and specific signal processing with source separation. CineJENSE reconstruction, based on implicit neural representations was adapted to free-breathing data. Correlation coefficient between both respiration signals and F1-score of the cardiac peak detections were computed for quantitative results. The reconstructed images were visually inspected to assess their quality and presence of motion artefacts and an automatic segmentation was performed and compared to the manual segmentation with DICE scores computation.Main results. An average correlation coefficient of 0.69 ± 0.22 and F1-score of 0.73 ± 0.23 for all subjects was found. Reconstructed images quality was close to that of the reconstructed images with reference signals, although slightly lower (2.51 ± 0.8 and 2.84 ± 0.7). Dice scores for LV was 0.86 ± 0.13 for reconstructed images with sensor-free estimations compared to 0.85 ± 0.12 with external sensors.Significance. This study demonstrated overall good quality images of free-breathing acquisitions using cardiac and respiration motion estimations based on the RF noise navigator.

Objective.Electrocardiograms (ECGs) contain valuable information in the clinical diagnosis of myocardial infarction (MI). However, its interpretation process is dependent on cardiologists with extensive clinical experience and expertise. The issue not only causes a paucity of medical resources, but also restricts patients from receiving timely diagnoses. Thus, a novel approach for MI automatic detection is developed, based on 12-lead ECG and an improved state refinement for long short-term memory (LSTM) determined 3D convolution-attention (3D CAISR-LSTM) model.Approach.The proposed 3D CAISR-LSTM model is trained in an end-to-end fashion. The input 12-lead ECG signals are preprocessed to eliminate power line interference, high-frequency noise and baseline wander. Then, the ECG signals are transformed into time-frequency images using continuous wavelet transform and bilinear interpolation. The obtained images are constructed into three-dimensional spatiotemporal features, serving as input to the 3D CAISR-LSTM model. In the 3D CAISR-LSTM model, there are three main components: a convolutional module, four identical convolutional attention modules, and an improved state refinement for LSTM. Performance of the 3D CAISR-LSTM model in automatic detection of MI versus healthy controls is evaluated through ten-fold cross validation on the publicly available PTB diagnostic ECG database.Main results.Experimental results demonstrate that the 3D CAISR-LSTM model achieves an accuracy of 98.45%, sensitivity of 98.69%, specificity of 97.50%, andF1 score of 99.03%, outperforming various advanced 2D and 3D deep neural network architectures.Significance.The proposed approach is expected to provide an early warning before obvious MI symptoms appear. It also has the potential to be developed into a lightweight embedded MI detection equipment.

Objective: Instantaneous determination of left ventricular (LV) diastolic function would be a useful aid in diagnosis and treatment of heart failure. The time constant of LV pressure decay (also known as Tau) is an established metric for evaluating LV stiffness and assessing LV diastolic function. Approach: In this study, we present a novel approach that uses a single arterial (aortic) pressure waveform to classify abnormal Tau through a physics-based machine learning (ML) methodology. This study is based on a clinical LV catheterization at the University of Southern California Keck Medical Center. We included 54 patients (13 females, age 36-90 (66.3±10.8), BMI 20.2-38.5 (27.8±4.6)) that were subjected to the same exclusion criteria of the primary study. Invasive pressure waveforms at LV and ascending aorta were measured using 2.5 F transducer tipped electronic micro-catheters. Intrinsic frequency (IF) parameters were computed from aortic pressure waveforms. Tau was calculated using an exponential curve-fitting approach based on LV pressure. Tau ranges were 25.7-86.5 ms (50.3±11), and Tau = 48 ms was used as a binary classification cut-off. Random forest and K-nearest neighbors classifiers were trained on 44 patients and blindly tested on 10 patients. 3- fold cross-validation was used to prevent overfitting. Main Results: Our proposed ML classifier model accurately predicts true Tau classes using physics-based features, where the most accurate one demonstrates 80.0% (elevated) and 100.0% (normal) success in predicting true Tau classes on blind data. Significance: We demonstrate that our proposed physics-based ML models can instantaneously classify Tau using information from a single aortic pressure waveform. Although an invasive proof, the required model inputs can be acquired non-invasively using carotid waveforms, working toward a smartphone-based, patient-activated tool for assessing diastolic dysfunction. .

Objective.The left ventricle (LV) pressure-volume (PV) loop provides comprehensive characteristic information into ventricular mechanics, aiding in the assessment of systolic and diastolic function. However, its routine clinical application is limited due to the invasiveness of conventional LV catheterization procedures. This study introduces a novel analytical framework for estimating LV pressure (LVP) waveforms noninvasively, using carotid pressure waveforms and routine cardiac imaging.Approach.The proposed method consists of a five-step analytical approach that integrates physical and physiological LV-aortic coupling relationships with a novel ventricular filling model. To assess the sensitivity and effectiveness of our method, we applied it on a clinical sample of 77 people (42% female), comprising healthy volunteers and heart failure (HF) patients, and analyzed the reconstructed PV-loops for key hemodynamic metrics.Main results.The proposed method robustly captured key hemodynamic changes associated with HF patients, including elevated LV end-diastolic pressure (p< 0.01), loss of inotropy (p< 0.001), and impaired ventricular efficiency (p< 0.001). Additionally, HF patients exhibited significantly smaller stroke work (p< 0.001), mean external power (p< 0.01), and contractility (p< 0.001) compared to the control group. These results align well with established clinical observations for HF, demonstrating the method's ability to detect pathological ventricular modifications.Significance.The proposed noninvasive LVP estimation method provides physiologically and clinically relevant PV-loop metrics without requiring invasive catheterization. By reliably capturing ventricular dysfunction in HF patients, this approach offers a promising alternative for noninvasive cardiac assessment. Its ability to enable routine evaluation of LV mechanics has the potential to improve HF diagnosis and therapeutic management, facilitating earlier intervention and more personalized treatment strategies.

Objective.The potential of optical pumped magnetometer magnetocardiography (OPM-MCG) for diagnosing coronary artery disease (CAD) has been initially shown, yet lacks large-scale prospective research.Approach.Using invasive coronary angiography (ICA) as a reference, we constructed three feature sets for the development of machine learning (ML) models: a 'Heart' feature set consisting only of OPM-MCG features, a 'Clinical' feature set, and a 'Heart + Clinical' combined feature set. We assessed the performance of 11 ML models with 10-fold cross-validation and conducted a feature importance analysis.Main results and Significance. Among 1513 participants (mean age 58.2 ± 12.0 years, 75.5% male), 1194 (78.92%) tested positive for ICA. Significant differences were observed in 'Heart' and 'Clinical' features between ICA-positive and negative groups. ML models using only 'Heart' features (AUC 0.84-0.88) outperformed those using only 'Clinical' features (AUC 0.62-0.75). Combining both feature types improved diagnostic accuracy (AUC 0.75-0.90). Feature importance analysis highlighted that 'Significant change of Ar-PN' in OPM-MCG was key for ICA diagnosis (47.8%), along with 'Abnormal Sp-TT', 'Significant change of Ps-PN', and 'Abnormal Mg-TT'. OPM-MCG has high performance in diagnosing CAD, and the significant change of Ar-PN is the most important feature. Cat Boost and random forest are more suitable for OPM-MCG to build ML diagnostic models for CAD.

Objectives.This study aimed to develop and validate a load cell-based device for measuring isometric forearm rotation torque and to determine its test-retest reliability.Approach.The custom-built device was calibrated using known weights and validated against a high-precision torque transducer. For reliability assessment, 35 physically active participants (20 males, 15 females; age 30 ± 7 years) were tested for isometric forearm pronation and supination strength 5-7 d apart.Main results.The custom device demonstrated excellent validity (intraclass correlation coefficient (ICC), absolute agreement = 1.00;r²= 1.00,p< 0.001; mean difference = -1.26-1.44%,p< 0.001). Test-retest reliability was excellent for absolute pronation and supination torque (ICC = 0.88-0.97; coefficient of variation percentage (CV%) = 4.1-5.6; minimal detectable change (MDC) at 90% confidence level = 13.1-19.9%), good to excellent for supination:pronation ratios (ICC = 0.60-0.88; CV% = 7.0-8.6; MDC = 0.10-0.13), and fair to good for dominant:non-dominant ratios (ICC = 0.42-0.66; CV% = 6.1-7.6; MDC = 0.07-0.10). Sex significantly influenced absolute torque values, with males demonstrating consistently higher torque, although reliability metrics were similar for both sexes.Significance.The device is valid, and the test is reliable. It is suitable for clinical assessments, rehabilitation monitoring, and performance evaluation, facilitating an improved understanding of factors affecting elbow overloading and injuries. Limb ratio metrics should be interpreted with caution due to their lower reliability.

Objective.Various time domain features, including dicrotic notch (dic), in photoplethysmogram (PPG), and the pulse transit time (PTT) determined using the simultaneously recorded electrocardiogram (ECG), are believed to have a critical role with many potential clinical applications. However, the dependence of these parameters on PPG sensor location is less well known.Approach.Three transmissive pulse oximetry probes (Xhale) were put simultaneously on the ear, nose, and finger of 36 healthy volunteers in the lower body negative pressure (LBNP) experiment. Various features of the recorded PPG signals were analyzed across different LBNP phases for each location. Simultaneously recorded finger PPG and ECG (Nellcor) were used to assess the dependence of PTT on PPG sensor location.Main results.PPG signal quality varies by measurement site, with nasal PPG showing the highest quality and ear PPG the lowest. Except pulse rate (PR), most feature-related indices differ across sites. Specifically, the ratios of detectabledicvary, highest in finger PPG and lowest in nasal PPG. Whendicis detectable, theepoint anddicare significantly different. PR variability indices and PTT also vary by location, though no clear conclusions can be drawn about PTT behavior across different LBNP phases.Significance.Various indices derived from PPG signals in a well-controlled study environment are influenced by sensor placement. Although not all possible indices are examined, the findings clearly illustrate the sensitivity of signal features to measurement location. While these results may not be directly generalizable to routine clinical settings, caution is warranted when extrapolating findings from one PPG site to another. This consideration is especially important in the digital health era, where mobile devices with PPG sensors are increasingly deployed at diverse body sites.

Objective: Depression is a prevalent mental health disorder that significantly impacts well-being and quality of life. This study investigates the relationship between depression and cardiovascular function, exploring time-series features derived from electrocardiogram (ECG) and photoplethysmogram (PPG) data as potential biomarkers for depression prescreening. Approach: As part of a comprehensive psycholinguistic experiment, we collected data from 60 individuals, including both healthy participants and those with varying levels of depression, assessed using the Beck Depression Inventory-II (BDI-II) and the Patient Health Questionnaire-9 (PHQ-9). Bimodal features derived from both ECG and PPG data were used to develop machine learning models for depression risk classification, employing classifiers such as Random Forest, XGBoost, Logistic Regression, and Support Vector Machines (SVM). Additionally, regression models were built to predict depression severity based on ECG- and PPG-derived biomarkers. Main Results: Key findings indicate that short-term variability (SD1) features in the ECG RR interval, peripheral systolic and diastolic phases from the PPG, and pulse duration significantly differ between healthy individuals and those at risk of depression. SVM achieved the best classification performance, with an AUROC of 0.83 ± 0.11 for BDI-II-based classification and 0.78 ± 0.11 for PHQ-9-based classification. SHAP analysis consistently identified systolic-SD1 and RR-SD1 as key predictors. Regression analysis further supported the role of cardiovascular features in assessing depression severity, yielding a mean absolute error (MAE) of 10.18 for BDI-II and 5.27 for PHQ-9 score regression. Significance: This study demonstrates the feasibility of using wearable ECG and PPG technologies for depression prescreening. The findings suggest that cardiac activity-based biomarkers can contribute to the development of cost-effective, objective, and non-invasive tools for mental health assessment, complementing traditional diagnostic methods.

Objectives.To quantify and evaluate the dynamic response of RR intervals (RRI) and heart rate (HR) measurements of commercially available Bluetooth chest-worn HR monitors during induced rapid changes in HR.Approach.An arbitrary function generator created synthetic electrocardiogram signals simulating the heart activity. Different scenarios of rapid changes in HR were simulated several times using: (1) step responses; (2) exercise data (EX); and (3) intermittent EX data. RRI and HR were recorded using the standard Bluetooth HR service for four wearable monitors: Garmin HRM-Dual, Movesense active, Polar H10, and Wahoo TRACKR. RRI latency, HR latency, and agreement were evaluated from the reference signal.Main results.RRI latency (median and interquartile range) was 0.7(0.5,0.7) s for Garmin, 0.4(0.2,0.5) s for Movesense, 2.6(2.2,2.8) s for Polar, and 2.1(1.9,2.4) s for Wahoo, where results did not differ greatly between tests. HR response latency was different between devices and tests. During intermittent EX tests, HR latency was 3.3(3.0, 3.3) s for Garmin, 1.0(1.0,1.0) s for Movesense, 2.3(2.3,2.3) s for Polar, and 2.2(2.2,2.3) s for Wahoo, where all devices consistently underestimated HR peaks and overestimated HR valleys, with a greater discrepancy in HR valleys.Significance.Most validation protocols of RRI and HR measured by wearable monitors neglect their dynamic characteristics. The present study demonstrated that manufacturers implemented different digital filters to compute the HR values, limiting the devices' ability to capture rapid HR changes. Open documentation of the processing steps is advised, and use cases involving sharp HR changes-such as intermittent high-intensity training-should rely on beat-to-beat RRI recordings.