Table of contents

Fetal monitoring during labour currently fails to accurately detect acidemia. We developed a method to assess the multidimensional properties of fetal heart rate variability (fHRV) from trans-abdominal fetal electrocardiogram (fECG) during labour. We aimed to assess this novel bioinformatics approach for correlation between fHRV and neonatal pH or base excess (BE) at birth.

We enrolled a prospective pilot cohort of uncomplicated singleton pregnancies at 38–42 weeks' gestation in Milan, Italy, and Liverpool, UK. Fetal monitoring was performed by standard cardiotocography. Simultaneously, with fECG (high sampling frequency) was recorded. To ensure clinician blinding, fECG information was not displayed. Data from the last 60 min preceding onset of second-stage labour were analyzed using clinically validated continuous individualized multiorgan variability analysis (CIMVA) software in 5 min overlapping windows. CIMVA allows simultaneous calculation of 101 fHRV measures across five fHRV signal analysis domains. We validated our mathematical prediction model internally with 80:20 cross-validation split, comparing results to cord pH and BE at birth.

The cohort consisted of 60 women with neonatal pH values at birth ranging from 7.44 to 6.99 and BE from −0.3 to −18.7 mmol L⁻¹. Our model predicted pH from 30 fHRV measures (R² = 0.90, P < 0.001) and BE from 21 fHRV measures (R² = 0.77, P < 0.001).

Novel bioinformatics approach (CIMVA) applied to fHRV derived from trans-abdominal fECG during labor correlated well with acid-base balance at birth. Further refinement and validation in larger cohorts are needed. These new measurements of fHRV might offer a new opportunity to predict fetal acid-base balance at birth.

Variability analysis of respiratory waveforms has been shown to provide key insights into respiratory physiology and has been used successfully to predict clinical outcomes. The current standard for quality assessment of the capnogram signal relies on a visual analysis performed by an expert in order to identify waveform artifacts. Automated processing of capnograms is desirable in order to extract clinically useful features over extended periods of time in a patient monitoring environment. However, the proper interpretation of capnogram derived features depends upon the quality of the underlying waveform. In addition, the comparison of capnogram datasets across studies requires a more practical approach than a visual analysis and selection of high-quality breath data. This paper describes a system that automatically extracts breath-by-breath features from capnograms and estimates the quality of individual breaths derived from them. Segmented capnogram breaths were presented to expert annotators, who labeled the individual physiological breaths into normal and multiple abnormal breath types. All abnormal breath types were aggregated into the abnormal class for the purpose of this manuscript, with respiratory variability analysis as the end-application. A database of 11 526 breaths from over 300 patients was created, comprising around 35% abnormal breaths. Several simple classifiers were trained through a stratified repeated ten-fold cross-validation and tested on an unseen portion of the labeled breath database, using a subset of 15 features derived from each breath curve. Decision Tree, K-Nearest Neighbors (KNN) and Naive Bayes classifiers were close in terms of performance (AUC of 90%, 89% and 88% respectively), while using 7, 4 and 5 breath features, respectively. When compared to airflow derived timings, the 95% confidence interval on the mean difference in interbreath intervals was ± 0.18 s. This breath classification system provides a fast and robust pre-processing of continuous respiratory waveforms, thereby ensuring reliable variability analysis of breath-by-breath parameter time series.

Though portable accelerometers are ubiquitous in physiology and public health studies, their accuracy as objective measures of physical activity is still being examined. This paper enumerates and analyzes the various biases of the widely used ActiLife^® software in reporting activity counts from ActiGraph^® accelerometers. In particular, we focus on the two-stage proprietary filtration algorithm used to convert raw acceleration data, for a sampling rate of 30 Hz, to compressed 1 Hz signals; we develop simple novel methods to analyze the action of the software filter on the raw data in the frequency domain.

The presence of motion artifacts in photoplethysmographic (PPG) signals is one of the major obstacles in the extraction of reliable cardiovascular parameters in continuous monitoring applications. In the current paper we present an algorithm for motion artifact detection based on the analysis of the variations in the time and the period domain characteristics of the PPG signal. The extracted features are ranked using a normalized mutual information feature selection algorithm and the best features are used in a support vector machine classification model to distinguish between clean and corrupted sections of the PPG signal. The proposed method has been tested in healthy and cardiovascular diseased volunteers, considering 11 different motion artifact sources. The results achieved by the current algorithm (sensitivity—SE: 84.3%, specificity—SP: 91.5% and accuracy—ACC: 88.5%) show that the current methodology is able to identify both corrupted and clean PPG sections with high accuracy in both healthy (ACC: 87.5%) and cardiovascular diseases (ACC: 89.5%) context.

Shoulder disorders, including rotator cuff tears, affect the shoulder function and result in adapted muscle activation. Although these adaptations have been studied in controlled conditions, free-living activities have not been investigated. Based on the kinematics measured with inertial sensors and portable electromyography, the objectives of this study were to quantify the duration of the muscular activation in the upper trapezius (UT), medial deltoid (MD) and biceps brachii (BB) during motion and to investigate the effect of rotator cuff tear in laboratory settings and daily conditions. The duration of movements and muscular activations were analysed separately and together using the relative time of activation (T_EMG/mov). Laboratory measurements showed the parameter's reliability through movement repetitions (ICC > 0.74) and differences in painful shoulders compared with healthy ones (p < 0.05): longer activation for UT; longer activation for MD during abduction and tendency to shorter activation in other movements; shorter activation for BB. In daily conditions, T_EMG/mov for UT was longer, whereas it was shorter for MD and BB (p < 0.05). Moreover, significant correlations were observed between these parameters and clinical scores. This study thus provides new insights into the rotator cuff tear effect on duration of muscular activation in daily activity.

The purposes of the present study were two fold: (1) to determine if the model used for estimating the physical working capacity at the fatigue threshold (PWC_FT) from electromyographic (EMG) amplitude data during incremental cycle ergometry could be applied to treadmill running to derive a new neuromuscular fatigue threshold for running, and (2) to compare the running velocities associated with the PWC_FT, ventilatory threshold (VT), and respiratory compensation point (RCP). Fifteen college-aged subjects (21.5  ±  1.3 y, 68.7  ±  10.5 kg, 175.9  ±  6.7 cm) performed an incremental treadmill test to exhaustion with bipolar surface EMG signals recorded from the vastus lateralis. There were significant (p < 0.05) mean differences in running velocities between the VT (11.3  ±  1.3 km h⁻¹) and PWC_FT (14.0  ±  2.3 km h⁻¹), VT and RCP (14.0  ±  1.8 km h⁻¹), but not the PWC_FT and RCP. The findings of the present study indicated that the PWC_FT model could be applied to a single continuous, incremental treadmill test to estimate the maximal running velocity that can be maintained prior to the onset of neuromuscular fatigue. In addition, these findings suggested that the PWC_FT, like the RCP, may be used to differentiate the heavy from severe domains of exercise intensity.

Our interest in the trabecular alignment within bone stems from the need to better understand the manner in which it can affect ultrasound propagation, particularly in pedicles. Within long bones it is well established that trabecular structures are aligned in an organized manner associated with the direction of load distribution; however, for smaller bones there are limited alignment studies. To investigate the directionality distribution in a quantitative manner we used a micro-CT to obtain three-dimensional (3D) structural data and developed analytical methods based on the special properties of Gabor filters. Implementation of these techniques has been developed and tested on a variety of simulated images as well as on 3D structures whose geometry is well-defined. To test the use of this technique we compared the results obtained on vertebral body trabecular bone with visual directionality and previous measurements by others. The method has been applied to six human pedicle samples in two orthogonal planes with results that provide reasonable proof-of-principle evidence that the method is well suited for estimating the directionality distribution within pedicle bones.

Very short-term heart rate variability (HRV) is thought to reflect dynamic changes in autonomic nervous activity, which is helpful in understanding the role of autonomic nervous function (ANF) in the mechanisms underlying apnea-induced cardiac arrhythmias. The goal of this study was to investigate the effect of repetitive end-inspiration breath holding on very short-term HRV. A total of 32 young healthy participants took part in the experiments. Three trials were performed, each involving seven repetitive end-inspiration breath holding and a 30 s recovery period between breath holding. Durations of breath holding in the three trials were 1:2:3. The study first evaluated the effect of analyzed data lengths on the stability of HRV indices and determined three HRV indices suitable for very short-term analysis. The results showed that in most cases, during breath holding, the square root of the mean squared differences of successive normal RR intervals (rMSSD) was significantly lower, but normalized units of the power in the low frequency band ranging from 0.04 to 0.15 Hz (nLF) and LF/high frequency (HF) were significantly higher than those during corresponding durations under the normal breathing conditions. On the contrary, during recovery after breath holding, rMSSD was significantly higher but nLF and LF/HF were lower than normal. Moreover, the durations of breath holding had no significant influence on the variations of LF/HF. In addition, as participants repeated the breath holding, HRV indices varied non-linearly. HRV changes may indicate sympathetic activation during breath holding and parasympathetic activation during recovery after breath holding. In conjunction with the existing physiological interpretation based on changes in heart rate, the results may imply that breath holding leads to both cardiac sympathetic and parasympathetic activation simultaneously, which may be a possible pathogenic factor of apnea-induced arrhythmias.

Nasal expiratory resistive valves (Provent^®) have been proposed as novel therapy for obstructive sleep apnea. We compared pressure measurements from a standard nasal pressure catheter used to assess nasal airflow during sleep with those from nasal expiratory resistive device with attached proprietary nasal pressure cannula. Nasal pressure cannula or Provent^® + proprietary nasal pressure cannula were attached to a bench model of human anterior nares and nasal passages, and pressure measured (P). Respiratory airflows generated by a subject breathing were applied to rear of model and airflow (${\rm \dot{V}}$) measured via pneumotachograph. Airflow amplitude (Δ${\rm \dot{V}}$) was plotted against pressure amplitude (ΔP). Hypopnoea detection (<50% Δ${\rm \dot{V}}$) sensitivity and specificity was tested by expressing ΔP in terms of two reference breaths: reference breath 1, Δ${\rm \dot{V}}$ 0.55 L s⁻¹ = 100%; and reference breath 2, Δ${\rm \dot{V}}$ 0.45 L s⁻¹ = 100%. ΔP/Δ${\rm \dot{V}}$ relationships were linear for Δ${\rm \dot{V}}$ ≤ 0.55 L s⁻¹; ΔP = 0.37ΔV + 0.16 (nasal pressure cannula), ΔP = 2.7ΔV + 0.12 (Provent^® + proprietary nasal pressure cannula); both R² > 0.65, p < 0.0001; p < 0.0001 for between slope difference). For nasal pressure cannula, specificity of hypopnoea detection differed between reference breaths one and two (80.2% and 40.0%, respectively), and Provent^® + proprietary nasal pressure cannula (30.3% and 74.2%, respectively). Quantification of airflow obstruction in the presence of Provent^® + proprietary nasal pressure cannula is greatly influenced by the reference breath chosen to determine a reduction in nasal airflow. Reported variability in therapeutic response to nasal expiratory resistive devices may relate to differences in measurement technique specificity used to quantify the severity of sleep disordered breathing.

Conventional analysis of clinical resting electroencephalography (EEG) recordings typically involves assessment of spectral power in pre-defined frequency bands at specific electrodes. EEG is a potentially useful technique in drug development for measuring the pharmacodynamic (PD) effects of a centrally acting compound and hence to assess the likelihood of success of a novel drug based on pharmacokinetic–pharmacodynamic (PK–PD) principles. However, the need to define the electrodes and spectral bands to be analysed a priori is limiting where the nature of the drug-induced EEG effects is initially not known. We describe the extension to human EEG data of a generalised semi-linear canonical correlation analysis (GSLCCA), developed for small animal data. GSLCCA uses data from the whole spectrum, the entire recording duration and multiple electrodes. It provides interpretable information on the mechanism of drug action and a PD measure suitable for use in PK–PD modelling. Data from a study with low (analgesic) doses of the μ-opioid agonist, remifentanil, in 12 healthy subjects were analysed using conventional spectral edge analysis and GSLCCA. At this low dose, the conventional analysis was unsuccessful but plausible results consistent with previous observations were obtained using GSLCCA, confirming that GSLCCA can be successfully applied to clinical EEG data.

Hydrogen sulfide (H₂S) is a toxic gas. It has been recognized that H₂S evolving in biochemical reactions in living organisms has an important role in different physiologic processes. Nowadays, H₂S is known as an endogenous messenger molecule. Natural sulfurous spring water has been proved beneficial in the therapy of diseases of the skin and other organs (Boros et al2013). In vivo real-time detection of local H₂S concentration is an important but challenging task.

We developed a two-electrode amperometric cell for selective subcutaneous detection of H₂S in anesthetized mice. The cell is a small size implantable gas sensor containing a platinum disc anode and a silver cathode. The selectivity is provided by a membrane permeable only by gases. There is a buffered reversible electrochemical mediator solution in an oxidized form inside the cell. As gaseous H₂S penetrates into the cell the mediator is reduced, and +0.4 V versus the reference is employed on the platinum working electrode. The reduced mediator is oxidized on the anode surface. The current provides an analytical signal representing the concentration of H₂S.

Appropriate shape, size and membrane material were selected, and optimal working parameters—such as mediator concentration, pH and cell voltage—were determined in vitro. The lower limit of detection in the stirred sample solution at pH = 5.5 was as small as 9.4  ×  10⁻⁷ M and a dynamic concentration range of 0–6  ×  10^–4 M could be achieved.

The detecting surfaces of the cell were covered with freshly dissected mouse skin to test dermal H₂S permeability. In other experiments, the cell was implanted subcutaneously in an anesthetized mouse and the animal was submerged in a buffer solution containing different concentrations of H₂S so that the skin surface over the sensor was covered by the solution. Measurements of subcutaneous H₂S concentration were taken. The experiments clearly proved that H₂S diffuses through the skin of the live mouse.

Snore analysis techniques have recently been developed for sleep studies. Most snore analysis techniques require reliable methods for the automatic classification of snore and breathing sounds in the sound recording. In this study we focus on this problem and propose an automated method to classify snore and breathing sounds based on the novel feature, 'positive/negative amplitude ratio (PNAR)', to measure the shape of the sound signal. The performance of the proposed method was evaluated using snore and breathing recordings (snore: 22 643 episodes and breathing: 4664 episodes) from 40 subjects. Receiver operating characteristic (ROC) analysis showed that the proposed method achieved 0.923 sensitivity with 0.918 specificity for snore and breathing sound classification on test data. PNAR has substantial potential as a feature in the front end of a non-contact snore/breathing-based technology for sleep studies.

Progressive narrowing of the upper airway increases airflow resistance and can produce snoring sounds and apnea/hypopnea events associated with sleep-disordered breathing due to airway collapse. Recent studies have shown that acoustic properties during snoring can be altered with anatomic changes at the site of obstruction. To evaluate the instantaneous association between acoustic features of snoring and the anatomic sites of obstruction, a novel method was developed and applied in nine patients to extract the snoring sounds during sleep while performing dynamic magnetic resonance imaging (MRI). The degree of airway narrowing during the snoring events was then quantified by the collapse index (ratio of airway diameter preceding and during the events) and correlated with the synchronized acoustic features. A total of 201 snoring events (102 pure retropalatal and 99 combined retropalatal and retroglossal events) were recorded, and the collapse index as well as the soft tissue vibration time were significantly different between pure retropalatal (collapse index, 24  ±  11%; vibration time, 0.2  ±  0.3 s) and combined (retropalatal and retroglossal) snores (collapse index, 13  ±  7% [P ≤ 0.0001]; vibration time, 1.2  ±  0.7 s [P ≤ 0.0001]). The synchronized dynamic MRI and acoustic recordings successfully characterized the sites of obstruction and established the dynamic relationship between the anatomic site of obstruction and snoring acoustics.

Nocturnal respiration rate parameters were collected from 20 COPD subjects over an 8 week period, to determine if changes in respiration rate were associated with exacerbations of COPD. These subjects were primarily GOLD Class 2 to 4, and had been recently discharged from hospital following a recent exacerbation. The respiration rates were collected using a non-contact radio-frequency biomotion sensor which senses respiratory effort and body movement using a short-range radio-frequency sensor. An adaptive notch filter was applied to the measured signal to determine respiratory rate over rolling 15 s segments. The accuracy of the algorithm was initially verified using ten manually-scored 15 min segments of respiration extracted from overnight polysomnograms. The calculated respiration rates were within 1 breath min⁻¹ for >98% of the estimates. For the 20 subjects monitored, 11 experienced one or more subsequent exacerbation of COPD (ECOPD) events during the 8 week monitoring period (19 events total). Analysis of the data revealed a significant increase in nocturnal respiration rate (e.g. >2 breath min⁻¹) prior to many ECOPD events. Using a simple classifier of a change of 1 breath min⁻¹ in the mode of the nocturnal respiration rate, a predictive rule showed a sensitivity of 63% and specificity of 85% for predicting an exacerbation within a 5 d window. We conclude that it is possible to collect respiration rates reliably in the home environment, and that the respiration rate may be a potential indicator of change in clinical status.

Polysomnography (PSG) has been extensively studied for sleep staging, where sleep stages are usually classified as wake, rapid-eye-movement (REM) sleep, or non-REM (NREM) sleep (including light and deep sleep). Respiratory information has been proven to correlate with autonomic nervous activity that is related to sleep stages. For example, it is known that the breathing rate and amplitude during NREM sleep, in particular during deep sleep, are steadier and more regular compared to periods of wakefulness that can be influenced by body movements, conscious control, or other external factors. However, the respiratory morphology has not been well investigated across sleep stages. We thus explore the dissimilarity of respiratory effort with respect to its signal waveform or morphology. The dissimilarity measure is computed between two respiratory effort signal segments with the same number of consecutive breaths using a uniform scaling distance. To capture the property of signal morphological dissimilarity, we propose a novel window-based feature in a framework of sleep staging. Experiments were conducted with a data set of 48 healthy subjects using a linear discriminant classifier and a ten-fold cross validation. It is revealed that this feature can help discriminate between sleep stages, but with an exception of separating wake and REM sleep. When combining the new feature with 26 existing respiratory features, we achieved a Cohen's Kappa coefficient of 0.48 for 3-stage classification (wake, REM sleep and NREM sleep) and of 0.41 for 4-stage classification (wake, REM sleep, light sleep and deep sleep), which outperform the results obtained without using this new feature.

Continuous cardiac monitoring of healthy and unhealthy patients can help us understand the progression of heart disease and enable early treatment. Optical pulse sensing is an excellent candidate for continuous mobile monitoring of cardiovascular health indicators, but optical pulse signals are susceptible to corruption from a number of noise sources, including motion artifact. Therefore, before higher-level health indicators can be reliably computed, corrupted data must be separated from valid data. This is an especially difficult task in the presence of artifact caused by ambulation (e.g. walking or jogging), which shares significant spectral energy with the true pulsatile signal. In this manuscript, we present a machine-learning-based system for automated estimation of signal quality of optical pulse signals that performs well in the presence of periodic artifact. We hypothesized that signal processing methods that identified individual heart beats (segmenting approaches) would be more error-prone than methods that did not (non-segmenting approaches) when applied to data contaminated by periodic artifact. We further hypothesized that a fusion of segmenting and non-segmenting approaches would outperform either approach alone. Therefore, we developed a novel non-segmenting approach to signal quality estimation that we then utilized in combination with a traditional segmenting approach. Using this system we were able to robustly detect differences in signal quality as labeled by expert human raters (Pearson's r = 0.9263). We then validated our original hypotheses by demonstrating that our non-segmenting approach outperformed the segmenting approach in the presence of contaminated signal, and that the combined system outperformed either individually. Lastly, as an example, we demonstrated the utility of our signal quality estimation system in evaluating the trustworthiness of heart rate measurements derived from optical pulse signals.

The purpose of this study was to examine whether circulatory occlusion of the hand impacts on regional forearm muscle haemodynamics as determined by the near-infrared spectroscopy (NIRS) venous occlusion technique (NIRS-VOT). Twenty-five young, healthy participants (18 males and 7 females; 28  ±  4 years; 71  ±  7 kg) completed two experimental protocols that were performed on the dominant arm: (1) a series of five venous occlusion trials with a suprasystolic cuff (>260 mmHg) applied to the wrist and (2) five venous occlusion trials without hand-occlusion. Both protocols were performed twice in a counterbalanced manner. NIRS data were obtained from the flexor digitorum superficialis (FDS) muscle using a dual wavelength, continuous-wave spectrophotometer. FDS muscle blood flow (${{\dot{Q}}_{\text{FDS}}}$), vascular conductance (C_FDS), O₂ consumption (${\dot{V}} {{\text{o}}_{\text{2FDS}}}$), and venous O₂ saturation (SvO₂) were calculated from NIRS data during the initial 5 s of venous occlusion. Circulatory occlusion of the hand via wrist cuffing significantly (P < 0.05) reduced ${{\dot{Q}}_{\text{FDS}}}$ (–36  ±  23%), C_FDS (–37  ±  23%), ${\dot{V}} {{\text{o}}_{2\text{FDS}}}$ (–14  ±  31%) and SvO₂ (–14  ±  12%). These findings indicate that hand-occlusion, via wrist cuffing, adversely impacts on regional forearm haemodynamics as determined by the NIRS-VOT. Consequently, it is recommended that future investigators avoid hand-occlusion when using the NIRS-VOT to quantify spontaneous haemodynamics of regional forearm muscle.

A new Radiometer™ hemoximeter (ABL-80) has recently become available to measure carboxyhaemoglobin concentration for the optimized CO-rebreathing method (oCOR-method). Within the English Institute of Sport (EIS), hemoximeters are used in three different laboratories; therefore, precision and agreement of total haemoglobin mass (tHbmass) determination across sites is essential, and comparison to the previous model OSM-3 is desirable. Six male and one female (age 30  ±  6 years, body mass 78.1  ±  10.6 kg) undertook the oCOR-method. Venous blood (~2 ml) was sampled immediately before and at 7 min during the oCOR-method; with seven replicates from each time point simultaneously analysed on five different Radiometer™ hemoximeters [OSM-3(1), OSM-3(2), ABL-80(1), ABL-80(2) and ABL-80(3)]. There were no differences (p > 0.05) between Δ%HbCO or mean tHbmass analysed with five different hemoximeters (OSM-3(1): 886  ±  167 g; OSM-3(2): 896  ±  160 g: ABL-80(1): 904  ±  157 g; ABL-80(2): 906  ±  163 g: ABL-80(3): 906  ±  162 g). However, the Bland–Altman plot revealed that there was closer agreement between ABL-80 machines for tHbmass than for the OSM-3. The variance (i.e. % error) across replicate samples decreased as the number of samples increased, with the error derived from the 'worse-case' scenario (single samples) being 1.2 to 1.6 fold greater in the OSM-3 than the ABL-80. Although there were no differences in the average tHbmass measured with five different hemoximeters, the new ABL-80 were in better agreement with each other compared to the old OSM-3. Previously, five replicates were required to achieve a low error using the OSM-3; however, three replicates are sufficient with the ABL-80 model to produce an error of ≤ 1% in tHbmass.

We aimed to determine whether statistical significant differences exist between the sets of results obtained from two devices used in our department for measuring brainstem auditory evoked potentials (BAEPs) and somatosensory evoked potentials (SEPs). We obtained BAEP and median and posterior tibial nerve SEP values bilaterally in ten healthy subjects. The tests were performed on the same subject using two devices consecutively. The equipment consisted of a Nicolet Viking-IV (Nicolet, Madison, WI, USA) and a Viking Select (Viasys Healthcare, Madison, WI, USA), and the same recording electrodes and stimulator (auditory and electrical) were used without modifying any postural position of the subject. The stimuli and recording parameters were the same for both devices. We obtained 20 sets of data for each type of test. The Bland–Altman plots as well as the one-sample t-test or Wilcoxon signed rank test were used to compare data between the two groups of data sets. We found no significant differences between the sets of values obtained with the two devices. Our analysis indicates that the two devices are equal in recording all different variables of BAEP and SEP, which allows us to combine the BAEP and SEP data obtained from the two devices for follow-up studies involving quantitative statistical methods.

This study received institutional approval (protocol number PRAG-154/2013).