Arterial Blood Pressure Time Series Research Articles

Visualization of the Intracranial Pressure and Time Burden in Childhood Brain Trauma: What We Have Learned One Decade on With KidsBrainIT.

To validate the intracranial pressure (ICP) dose-response visualization plot for the first time in a novel prospectively collected pediatric traumatic brain injury (pTBI) data set from the multi-center, multi-national KidsBrainIT consortium. Prospectively collected minute-by-minute ICP and mean arterial blood pressure time series of 104 pTBI patients were categorized in ICP intensity-duration episodes. These episodes were correlated with the 6-month Glasgow Outcome Score (GOS) and displayed in a color-coded ICP dose-response plot. The influence of cerebrovascular reactivity and cerebral perfusion pressure (CPP) were investigated. The generated ICP dose-response plot on the novel data set was similar to the previously published pediatric plot. This study confirmed that higher ICP episodes were tolerated for a shorter duration of time, with an approximately exponential decay curve delineating the positive and negative association zones. ICP above 20 mm Hg for any duration in time was associated with poor outcome in our patients. Cerebrovascular reactivity state did not influence their respective transition curves above 10 mm Hg ICP. CPP below 50 mm Hg was not tolerated, regardless of ICP and duration, and was associated with worse outcome. The ICP dose-response plot was reproduced in a novel and independent pTBI data set. ICP above 20 mm Hg and CPP below 50 mm Hg for any duration in time were associated with worse outcome. This highlighted a pressing need to reduce pediatric ICP therapeutic thresholds used at the bedside.

Long-term prediction of blood pressure time series using ANFIS system based on DKFCM clustering

Long-term prediction of blood pressure time series could potentially help clinicians to take timely preventative steps even before life-threatening medical events such as an acute hypotensive episode (AHE). However, it is a challenging task to predict the time series in a long-term horizon due to its chaotic and nonlinear nature. In this regard, this paper firstly presents a comparative study on fuzzy clustering algorithms including Fuzzy C-Means (FCM), Kernel-based FCM (KFCM), Density-oriented KFCM (DKFCM), and Gustafson–Kessel (GK). After that, it introduces a new system for a 60-minute prediction of mean arterial blood pressure time series in ICU patients. This system consists of four steps. First, signal filtering is executed by using the empirical mode decomposition method. After that, the time series is embedded in dimension 4 and time delay 1 min. Then, a one-step prediction is made using an Adaptive Neuro-Fuzzy Inference System (ANFIS) based on DKFCM clustering (DKFCM-ANFIS). In the end, a multi-step prediction is made by a recursive strategy. We used three benchmark synthetic datasets viz D12, D15, and DUNN for clustering. Furthermore, PhysioNet/CinC challenge 2009 dataset was utilized for blood pressure prediction. Results show that DKFCM among the mentioned clustering algorithms achieved the best clustering performance. Compared to FCM-ANFIS, KFCM-ANFIS, and GK-ANFIS models, the proposed system had the longest prediction horizon as well as the least root mean square error in 57 out of 60 patients. Moreover, it demonstrated the highest accuracy, sensitivity, and specificity in predicting AHE 60 min ahead.

Chaos based nonlinear analysis to study cardiovascular responses to changes in posture

Cardiovascular responses to changes in posture have been the focus of numerous investigations in the past. Analysis of ECG and arterial blood pressure (ABP) data for different postures, obtained from ‘Physionet’ was done from a new perspective of analyzing nonlinear time series having multifractal features. Multifractal detrended fluctuation analysis (MFDFA) and also the cross-correlation between ECG and BP with another technique Multifractal detrended cross-correlation analysis (MFDXA) reveals ECG and ABP time series of different subjects to be multifractal in nature with different degree of complexity. Different degree of auto and cross-correlation for both ECG and ABP signals is also observed for all subjects.Further significant information obtained from the cross-correlation parameters obtained from MFDXA analysis is very important and relevant particularly for understanding the dynamics of orthostatic stress.

Occurrence of CPPopt Values in Uncorrelated ICP and ABP Time Series.

Optimal cerebral perfusion pressure (CPPopt) is a concept that uses the pressure reactivity (PRx)-CPP relationship over a given period to find a value of CPP at which PRx shows best autoregulation. It has been proposed that this relationship be modelled by a U-shaped curve, where the minimum is interpreted as being the CPP value that corresponds to the strongest autoregulation. Owing to the nature of the calculation and the signals involved in it, the occurrence of CPPopt curves generated by non-physiological variations of intracranial pressure (ICP) and arterial blood pressure (ABP), termed here "false positives", is possible. Such random occurrences would artificially increase the yield of CPPopt values and decrease the reliability of the methodology.In this work, we studied the probability of the random occurrence of false-positives and we compared the effect of the parameters used for CPPopt calculation on this probability. To simulate the occurrence of false-positives, uncorrelated ICP and ABP time series were generated by destroying the relationship between the waves in real recordings. The CPPopt algorithm was then applied to these new series and the number of false-positives was counted for different values of the algorithm's parameters. The percentage of CPPopt curves generated from uncorrelated data was demonstrated to be 11.5%. This value can be minimised by tuning some of the calculation parameters, such as increasing the calculation window and increasing the minimum PRx span accepted on the curve.

Collision frequency locality-sensitive hashing for prediction of critical events.

We present a fast, efficient method to predict future critical events for a patient. The prediction method is based on retrieving and leveraging similar waveform trajectories from a large medical database. Locality-sensitive hashing (LSH), our theoretical foundation, is a model-free, sub-linear time, approximate search method enabling a fast retrieval of a nearest neighbor set for a given query. We propose a new variant of LSH, namely Collision Frequency LSH (CFLSH), to further improve the prediction accuracy without sacrificing any speed. The key idea is that the more frequently an element and a query collide across multiple LSH hash tables, the more similar they are. Unlike the standard LSH which only utilizes the linear distance calculation, in CFLSH, the short-listing step from a pool of pre-selected candidates filtered by hash functions to the final nearest neighbor set relies upon the frequency of collision along with distance information. We evaluate CFLSH versus the standard LSH using the L1 and cosine distances, for predicting acute hypotensive episodes on arterial blood pressure time series data extracted from the MIMIC II database. Our results show that CFLSH for the L1 distance has a higher prediction accuracy and further accelerates the sub-linear querying time obtained by the standard LSH.

Stratified locality-sensitive hashing for accelerated physiological time series retrieval.

We introduce stratified locality-sensitive hashing (SLSH) for retrieving similar physiological waveform time series. SLSH further accelerates the sublinear retrieval time obtained by the standard locality-sensitive hashing (LSH) method. The standard family of locality-sensitive hash functions is limited to provide only a single perspective on the data due to its one-to-one relationship to a distinct distance function for measuring similarity. SLSH incorporates multiple locality-sensitive hash families with various distance functions enabling it to examine the data with more diverse and refined perspectives. We provide the procedures of SLSH with locality-sensitive hash families for the l1 and the cosine distances, and compare its performance to the standard LSH on an arterial blood pressure time series data extracted from the physiological waveform repository of the MIMIC2 database. The time to retrieve five most similar waveforms by SLSH is 14 times faster than the linear search and 1.7 times faster than the standard LSH when we allow 5% decrease in accuracy as a trade-off.

Hypotension Investigation, Prospective Clinical Study

Hypotension occurring in the initial phase of resuscitation is significantly associated with increased mortality following brain injury, even when episodes are relatively short. A large amount of data exists in health care systems providing information on the major health indicators of patients in hospitals. It is believed that if enough of these data could be drawn together and analysed in a systematic way, then a system could be built that will trigger an alarm predicting the onset of a hypotensive event. In the paper the mathematical information algorithm based on the concept of the rank of a sequence is presented. For the analysis of hypotension physiology an application of a new algebraic method is proposed for real world time series analysis. Numerical experiments with a hypotension crisis prevention using arterial blood pressure time series are used to illustrate the potential of the proposed method. The algorithm for finding ranks of a sequence of the ECG parameters is presented in the paper in order to show complexity profiles. Experimental results show that presented in this paper method also can be used together with other hypotension prediction methods. DOI: http://dx.doi.org/10.5755/j01.eie.22.2.13454

Large-scale physiological waveform retrieval via locality-sensitive hashing.

We propose a fast, scalable locality-sensitive hashing method for the problem of retrieving similar physiological waveform time series. When compared to the naive k-nearest neighbor search, the method vastly speeds up the retrieval time of similar physiological waveforms without sacrificing significant accuracy. Our result shows that we can achieve 95% retrieval accuracy or better with up to an order of magnitude of speed-up. The extra time required in advance to create the optimal data structure is recovered when query quantity equals 15% of the repository, while the method incurs a trivial additional memory cost. We demonstrate the effectiveness of this method on an arterial blood pressure time series dataset extracted from the ICU physiological waveform repository of the MIMIC-II database.

Cardiovascular responses after brisk finger movement and their dependency on the “eigenfrequency” of the baroreflex loop

The baroreflex is mainly involved in short-term blood pressure regulation and strongly influenced by activations of medullary circulation centres in the brain stem and higher brain centres. One important feature of the baroreflex is its strong preference for oscillations around 0.1 Hz, which can be seen as resonance or “eigenfrequency” (EF) of the control loop (so-called Mayer waves). In the present study we investigated beat-to-beat heart rate intervals (RRI) and arterial blood pressure (BP) changes after brisk finger movement and their relationship to the “eigenfrequency” determined by cross spectral analysis between RRI and arterial blood pressure time series of 17 healthy subjects. The analyses revealed significant correlations between BP response magnitude ( r = 0.63, p < 0.01) respectively RRI response magnitude ( r = 0.59, p < 0.05) and EF. This can be interpreted in such a way that subjects with a “high” EF (>0.10 Hz) elicit larger BP responses as well as larger RRI responses when compared to subjects with a “low” EF (<0.10 Hz).

Variability and Fractal Analysis of Middle Cerebral Artery Blood Flow Velocity and Arterial Blood Pressure in Subarachnoid Hemorrhage

Higher biologic systems operate far from equilibrium resulting in order, complexity, fluctuation of inherent parameters, and dissipation of energy. According to the decomplexification theory, disease is characterized by a loss of system complexity. We analyzed such complexity in patients after subarachnoid hemorrhage (SAH), by applying the standard technique of variability analysis and the novel method of fractal analysis to middle cerebral artery blood flow velocity (FV) and arterial blood pressure (ABP). In 31 SAH -patients, FV (using transcranial Doppler sonography) and direct ABP were measured. The standard deviations (s.d.) and coefficients of variation (CV=relative s.d.) for FV and ABP time series of length 2(10) secs were calculated as measures of variability. The spectral index beta(low) and the Hurst coefficient H(bdSWV) were analyzed as fractal measures. Outcome was assessed 1 year after SAH according to the Glasgow Outcome Scale (GOS). Both FV (beta(low)=2.2+/-0.4, mean+/-s.d.) and ABP (beta(low)=2.3+/-0.4) were classified as nonstationary (fractal Brownian motion) signals. FV showed significantly (P<0.05) higher variability (CV=7.2+/-2.5%) and Hurst coefficient (H(bdSWV)=0.26+/-0.13) as compared with ABP (CV=5.5+/-2.7%, H(bdSWV)=0.19+/-0.11). Better outcome (GOS) correlated significantly (P<0.05) with higher s.d. of FV (Spearman's r(s)=0.51, r(s)(2)=0.26) and ABP (r(s)=0.57, r(s)(2)=0.32), as well as with a higher Hurst coefficient of ABP (r(s)=0.46, r(s)(2)=0.21). Cerebral vasospasm reduced CV of FV, but left H(bdSWV) unchanged. FV and ABP fluctuated markedly despite homeostatic control. A reduced variability of FV and ABP might indicate a loss of complexity and was associated with a less favorable outcome. Therefore, the decomplexification theory of illness may apply to SAH.

Wavelet assessment of cerebrospinal compensatory reserve and cerebrovascular reactivity

We introduce a wavelet transfer model to relate spontaneous arterial blood pressure (ABP) fluctuations to intracranial pressure (ICP) fluctuations. We employ a complex continuous wavelet transform to develop a consistent mathematical framework capable of parametrizing both cerebral compensatory reserve and cerebrovascular reactivity. The frequency-dependent gain and phase of the wavelet transfer function are introduced because of the non-stationary character of the ICP and ABP time series. The gain characterizes the dampening of spontaneous ABP fluctuations and is interpreted as a novel measure of cerebrospinal compensatory reserve. For a group of 12 patients who died as a result of cerebral lesions (Glasgow Outcome Scale (GOS) = 1) the average gain in the low-frequency (0.02– 0.07 Hz) range was 0.51 ± 0.13 and significantly exceeded that of 17 patients with GOS = 2 having an average gain of 0.26 ± 0.11 with p = 1×10−4 (Kruskal–Wallis test). A time-averaged synchronization index (which may vary from 0 to 1) defined in terms of the wavelet transfer function phase yields information about the stability of the phase difference of the ABP and ICP signals and is used as a cerebrovascular reactivity index. A low value of synchronization index reflects a normally reactive vascular bed, while a high value indicates pathological entrainment of ABP and ICP fluctuations. Such entrainment is strongly pronounced in patients with fatal outcome (for this group the low-frequency synchronization index was 0.69 ± 0.17). The gain and synchronization parameters define a cerebral hemodynamic state space (CHS) in which the patients with GOS = 1 are to large extent partitioned away from those with GOS = 2. The concept of CHS elucidates the interplay of vascular and compensatory mechanisms.

Complexity and "chaos" in blood pressure after baroreceptor denervation of conscious dogs.

To investigate how arterial baroreceptors affect the dynamic properties of short-term blood pressure control, we determined Lyapunov exponents and correlation dimensions of blood pressure. Two groups of conscious dogs were studied: a control group (n = 7) and a group subjected to total sinoaortic and cardiopulmonary baroreceptor denervation (n = 7). As a measure of variability, standard deviation was determined and power spectra were calculated. In the lower frequency range (f < 0.1 Hz) power density was inversely related to frequency in both groups, indicating "1/f noise." Estimating the correlation dimension via the Grassberger-Procaccia algorithm as a quantification of complexity revealed a decrease after baroreceptor denervation (1.74 +/- 0.2 vs. 3.05 +/- 0.23 control; P < 0.05). Determination of the largest Lyapunov exponents lambda 1, which indicates the sensitive dependence on initial conditions, a hallmark of chaos, also yielded a diminution after denervation (lambda 1 = 0.74 +/- 0.08 vs. 1.85 +/- 0.18, P < 0.01). The results were cross-checked with surrogate data statistics. The null hypothesis, that there is no nonlinear structure in arterial blood pressure time series, was rejected. This shows that after baroreceptor denervation, blood pressure control is less complex and less sensitive to initial conditions ("chaos"). In contrast, variability (standard deviation) is increased (22.2 +/- 3.1 denervation vs. 8.3 +/- 1.4 control; P < 0.05). It is concluded that under physiological conditions, arterial and cardiopulmonary baroreceptors reduce variability of blood pressure, however, at the cost of blood pressure being less predictable. Thus the regulation is more sensitive depending on initial conditions.(ABSTRACT TRUNCATED AT 250 WORDS)