Study of multiparameter respiratory pattern complexity in surgical critically ill patients during weaning trials

Background  

Separation from mechanical ventilation is a difficult task, whereas conventional predictive indices have not been proven accurate enough, so far. A few studies have explored changes of breathing pattern variability for weaning outcome prediction, with conflicting results. In this study, we tried to assess respiratory complexity during weaning trials, using different non-linear methods derived from theory of complex systems, in a cohort of surgical critically ill patients.     

Sample-hold technique for analysis of respiratory gas composition at high breathing frequencies

A gas sampling device is described for continuous monitoring of respiratory gas composition that is applicable to experimental conditions when the breathing frequency is very high (greater than 2 Hz) and the response time of conventional gas analyzers becomes a critical limiting factor. The system utilizes the principle of discontinuous gas collection at any selected point of the respiratory cycle facilitated by ultraspeed piezoelectric valves and includes provision for sample-hold characteristics. Two distinct modes of operation are supported. In phase-locked mode gas sampling is synchronous with breathing frequency. In scanning mode gas collection is asynchronous with breathing frequency. Phase-locked mode may be used for continuous monitoring of end-tidal gas concentrations, whereas scanning mode is intended for assessing the gas concentration profile throughout the respiratory cycle. The system may be applied to steady breathing encountered in mechanical ventilation at high frequency or during quasi-steady breathing observed in panting animals. Combined with a respiratory mass spectrometer, the system has been used for measurement of gas concentrations in alveolar gas mixtures at breathing frequencies ranging from 3 to 30 Hz that were otherwise not amenable to rapid measuring techniques.     

Role of the autonomic nervous system in generating non-linear dynamics in short-term heart period variability.

We evaluated the role played by the autonomic nervous system in producing non-linear dynamics in short heart period variability (HPV) series recorded in healthy young humans. Non-linear dynamics are detected using an index of predictability based on a local non-linear predictor and a surrogate data approach. Different types of surrogates are utilized: (i) phase-randomized Fourier-transform based (FT) data; (ii) amplitude-adjusted FT (AAFT) data; and (iii) iteratively refined AAFT (IAAFT) data of two types (IAAFT-1 and IAAFT-2). The approach was applied to experimental protocols activating or blocking the sympathetic or parasympathetic branches of the autonomic nervous system or periodically perturbing cardiovascular control via paced respiration at different breathing rates. We found that short-term HPV was mostly linear at rest. Experimental protocols activating the sympathetic or parasympathetic nervous system did not produce non-linear dynamics. In contrast, paced respiration, especially at slow breathing rates, elicited significantly non-linear dynamics. Therefore, in short-term HPV ( approximately 300 beats) the use of non-linear models is not supported by the data, except under conditions whereby the subject is constrained to a slow respiratory rate.     

Patient-specific respiratory models using dynamic 3D MRI: Preliminary volunteer results

Organ and tumour motion has a significant impact on the planning and delivery of radiotherapy treatment. At present imaging modality such as four-dimensional computer tomography (4DCT) cannot be used to measure the variability of motion between different respiratory cycles. To create reliable motion models, one needs to acquire volumetric data sets of the lungs with sufficient sampling of the breathing cycle. In this paper we investigate the use of highly parallel MRI to acquire such data. A 32 channel coil in conjunction with a balanced SSFP sequence and a SENSE factor of 6 were used to acquire volumetric data sets in five healthy volunteers. The acquisition was repeated for seven series of different breathing patterns. The data acquired was of sufficient spatial resolution (5 × 5 × 5 mm³) and image quality to carry out automated non-rigid registration. The acquisition rate (c.a. 2 volumes per second) allowed for a meaningful sampling of the different respiratory curves that were automatically obtained from the skin surface motion. This acquisition technique should provide images of high enough quality to create statistical respiratory models.     

Genomic prediction based on data from three layer lines using non-linear regression models

Background

Most studies on genomic prediction with reference populations that include multiple lines or breeds have used linear models. Data heterogeneity due to using multiple populations may conflict with model assumptions used in linear regression methods.

Methods

In an attempt to alleviate potential discrepancies between assumptions of linear models and multi-population data, two types of alternative models were used: (1) a multi-trait genomic best linear unbiased prediction (GBLUP) model that modelled trait by line combinations as separate but correlated traits and (2) non-linear models based on kernel learning. These models were compared to conventional linear models for genomic prediction for two lines of brown layer hens (B1 and B2) and one line of white hens (W1). The three lines each had 1004 to 1023 training and 238 to 240 validation animals. Prediction accuracy was evaluated by estimating the correlation between observed phenotypes and predicted breeding values.

Results

When the training dataset included only data from the evaluated line, non-linear models yielded at best a similar accuracy as linear models. In some cases, when adding a distantly related line, the linear models showed a slight decrease in performance, while non-linear models generally showed no change in accuracy. When only information from a closely related line was used for training, linear models and non-linear radial basis function (RBF) kernel models performed similarly. The multi-trait GBLUP model took advantage of the estimated genetic correlations between the lines. Combining linear and non-linear models improved the accuracy of multi-line genomic prediction.

Conclusions

Linear models and non-linear RBF models performed very similarly for genomic prediction, despite the expectation that non-linear models could deal better with the heterogeneous multi-population data. This heterogeneity of the data can be overcome by modelling trait by line combinations as separate but correlated traits, which avoids the occasional occurrence of large negative accuracies when the evaluated line was not included in the training dataset. Furthermore, when using a multi-line training dataset, non-linear models provided information on the genotype data that was complementary to the linear models, which indicates that the underlying data distributions of the three studied lines were indeed heterogeneous.

Electronic supplementary material

The online version of this article (doi:10.1186/s12711-014-0075-3) contains supplementary material, which is available to authorized users.     

Testing prediction accuracy in short-term ecological studies

Applied ecology is based on an assumption that a management action will result in a predicted outcome. Testing the prediction accuracy of ecological models is the most powerful way of evaluating the knowledge implicit in this cause-effect relationship, however, the prevalence of predictive modeling and prediction testing are spreading slowly in ecology. The challenge of prediction testing is particularly acute for small-scale studies, because withholding data for prediction testing (e.g., via k-fold cross validation) can reduce model precision. However, by necessity small-scale studies are common. We use one such study that explored small mammal abundance along an elevational gradient to test prediction accuracy of models with varying degrees of information content. For each of three small mammal species, we conducted 5000 iterations of the following process: (1) randomly selected 75 % of the data to develop generalized linear models of species abundance that used detailed site measurements as covariates, (2) used an information theoretic approach to compare the top model with detailed covariates to habitat type-only and null models constructed with the same data, (3) tested those models’ ability to predict the 25 % of the randomly withheld data, and (4) evaluated prediction accuracy with a quadratic loss function. Detailed models fit the model-evaluation data best but had greater expected prediction error when predicting out-of-sample data relative to the habitat type models. Relationships between species and detailed site variables may be evident only within the framework of explicitly hierarchical analyses. We show that even with a small but relatively typical dataset (n = 28 sampling locations across 125 km over two years), researchers can effectively compare models with different information content and measure models’ predictive power, thus evaluating their own ecological understanding and defining the limits of their inferences. Identifying the appropriate scope of inference through prediction testing is ecologically valuable and is attainable even with small datasets.     

Lung and chest wall impedances in the dog: effects of frequency and tidal volume.

Dependences of the mechanical properties of the respiratory system on frequency (f) and tidal volume (VT) in the normal ranges of breathing are not clear. We measured, simultaneously and in vivo, resistance and elastance of the total respiratory system (Rrs and Ers), lungs (RL and EL), and chest wall (Rcw and Ecw) of five healthy anesthetized paralyzed dogs during sinusoidal volume oscillations at the trachea (50-300 ml, 0.2-2 Hz) delivered at a constant mean lung volume. Each dog showed the same f and VT dependences. The Ers and Ecw increased with increasing f to 1 Hz and decreased with increasing VT up to 200 ml. Although EL increased slightly with increasing f, it was independent of VT. The Rcw decreased from 0.2 to 2 Hz at all VT and decreased with increasing VT. Although the RL decreased from 0.2 to 0.6 Hz and was independent of VT, at higher f RL tended to increase with increasing f and VT (i.e., as peak flow increased). Finally, the f and VT dependences of Rrs were similar to those of Rcw below 0.6 Hz but mirrored RL at higher f. These data capture the competing influences of airflow nonlinearities vs. tissue nonlinearities on f and VT dependence of the lung, chest wall, and total respiratory system. More specifically, we conclude that 1) VT dependences in Ers and Rrs below 0.6 Hz are due to nonlinearities in chest wall properties, 2) above 0.6 Hz, the flow dependence of airways resistance dominates RL and Rrs, and 3) lung tissue behavior is linear in the normal range of breathing.     

Respiratory influences on non-linear dynamics of heart rate variability in humans

 The goal of our study was to determine whether evidence for chaos in heart rate variability (HRV) can be observed when the respiratory input to the autonomic controller of heart rate is forced by the deterministic pattern associated with periodic breathing. We simultaneously recorded, in supine healthy volunteers, RR intervals and breathing volumes for 20 to 30 min (1024 data point series) during three protocols: rest (control), fixed breathing (15 breath/min) and voluntary periodic breathing (3 breaths with 2 s inspiration and 2 s expiration followed by an 8 s breath hold). On both the RR interval and breathing volume series we applied the non-linear prediction method (Sugihara and May algorithm) to the original time series and to distribution-conserved isospectral surrogate data. Our results showed that, in contrast to the control test, during both fixed and voluntary periodic breathing the variability of breathing volumes was clearly deterministic non-chaotic. During all the three protocols, the RR-interval series’ non-linear predictability was consistent with one of a chaotic series. However, at rest, no clear difference was observed between the RR-interval series and their surrogates, which means that no clear chaos was observed. During fixed breathing a difference appeared, and this difference seemed clearer during voluntary periodic breathing. We concluded that HRV dynamics were chaotic when respiration was forced with a deterministic non-chaotic pattern and that normal spontaneous respiratory influences might mask the normally chaotic pattern in HRV. Received: 7 August 1995 / Accepted in revised form: 20 March 1997     

Using general linear model,Bayesian Networks and Naive Bayes classifier for prediction of Karenia selliformis occurrences and blooms

The prediction of the dinoflagellate red tide forming Karenia selliformis is a relevant task to aid optimized management decisions in marine coastal water. The objective of the present study is to compare different modeling approaches for prediction of Karenia selliformis occurrences and blooms. A set of physical parameters (salinity, temperature and tide amplitude), meteorological constraints (evaporation, air temperature, insolation, rainfall, atmospheric pressure and humidity), sampling months and sampling sites are used. The model prediction included general linear model (GLM), Bayesian Network (BN) and the simplest BN type which is, Naive Bayes classifier (NB). The results showed that three models incriminated high salinity in Karenia selliformis blooms and the sampling sites, mainly Boughrara lagoon, in the occurrences. The BN performed better than linear models (NB and GLM) for both Karenia selliformis occurrences and blooms prediction. This later is related to the facts that BN considered the inter-independency between predictive variables and that the relationships between the variables and the outcome are often non-linear such us; the transition to bloom situations appeared to be triggered by a salinity threshold. This study is useful in the management of this ecosystem so as to use the best disposal options in the early prediction of the toxic blooms.     

3D TOF-PET image reconstruction using total variation regularization

In this paper we introduce a semi-analytic algorithm for 3-dimensional image reconstruction for positron emission tomography (PET). The method consists of the back-projection of the acquired data into the most likely image voxel according to time-of-flight (TOF) information, followed by the filtering step in the image space using an iterative optimization algorithm with a total variation (TV) regularization. TV regularization in image space is more computationally efficient than usual iterative optimization methods for PET reconstruction with full system matrix that use TV regularization. The efficiency comes from the one-time TOF back-projection step that might also be described as a reformatting of the acquired data. An important aspect of our work concerns the evaluation of the filter operator of the linear transform mapping an original radioactive tracer distribution into the TOF back-projected image. We obtain concise, closed-form analytical formula for the filter operator. The proposed method is validated with the Monte Carlo simulations of the NEMA IEC phantom using a one-layer, 50 cm-long cylindrical device called Jagiellonian PET scanner. The results show a better image quality compared with the reference TOF maximum likelihood expectation maximization algorithm.     

Coordination of respiratory muscles assessed by means of nonlinear forecasting of demodulated myographic signals

Pulmonary diseases such as obstructive sleep apnea syndrome (OSAS) affect function of respiratory muscles. Individuals with OSAS suffer intermittent collapse of the upper airways during sleep due to unbalanced forces generated by the contraction of the diaphragm and upper airway dilator muscles.Respiratory rhythm and pattern generation can be described via nonlinear or coupled oscillators; therefore, the resulting activation of different respiratory muscles may be related to complex nonlinear interactions. The aims of this work were: to evaluate locally linear models for fitting and prediction of demodulated myographic signals from respiratory muscles; and to analyze quantitatively the influence of a pulmonary disease on this nonlinear forecasting related to low and moderate levels of respiratory effort.Electromyographic and mechanomyographic signals from three respiratory muscles (genioglossus, sternomastoid and diaphragm) were recorded in OSAS patients and controls while awake during an increased respiratory effort.Variables related to auto and cross prediction between muscles were calculated from the r² coefficient and the estimation of residuals, as functions of prediction horizon. In general, prediction improved linearly with higher levels of effort.A better prediction between muscle activities was obtained in OSAS patients when using genioglossus as the predictor signal. The prediction was significant for more than two respiratory cycles in OSAS patients compared to only a half cycle in controls. It could be concluded that nonlinear forecasting applied to genioglossus coupling with other muscles provides a promising assessment to monitor pulmonary diseases.     

Statistical methods for predicting the spatial abundance of reef fish species

Understanding the spatial distribution of organism abundance is fundamental to assessing and managing ecological populations. Marine species can be difficult and logistically challenging and expensive to observe. This often results in spatial data containing low detection rates when sampling underwater, biasing spatial predictions from many modeling approaches. We propose a multistage statistical workflow that can use zero inflated sampling data to develop non-linear predictive spatial distributions of reef fish abundance. The workflow includes: (1) an individual-based discrete event simulation which generates simulated survey data under different abundance settings; (2) empirical maximum likelihood analysis to establish the relationship between survey data and abundance from the simulation; (3) a two-step random smoothing method to estimate reliable block spatial abundance around each survey station; (4) an ensemble of different machine learning models which use the estimated abundance from step three as input to compute a stable non-linear prediction of abundance across the entire study area (Gulf of Mexico). Applying our workflow greatly improved the ability to forecast abundance at small spatial scales. The ability to forecast at fine spatial scales is critical when working with species that are patchily distributed. This workflow can apply to many ecological populations to develop abundance maps even if sample data is not well distributed across the study area or is zero inflated.     

Qualitative theory of compartmental systems with lags

Dynamic models of many processes in the biological and physical sciences give systems of ordinary differential equations called compartmental systems. Often, these systems include time lags; in this context, continuous probability density functions (pdfs) of lags are far more important than discrete lags. There is a relatively complete theory of compartmental systems without lags, both linear and non-linear [SIAM Rev. 35 (1993) 43]. The authors extend their previous work on compartmental systems without lags to show that, for discrete lags and for a very large class of pdfs of continuous lags, compartmental systems with lags are equivalent to larger compartmental systems without lags. Consequently, the properties of compartmental systems with lags are the same as those of compartmental systems without lags. For a very large class of compartmental systems with time lags, one can show that the time lags themselves can be generated by compartmental systems without lags. Thus, such systems can be partitioned into a main system, which is the original system without the lags, plus compartmental subsystems without lags that generate the lags. The latter may be linear or non-linear and may be inserted into main systems that are linear or non-linear. The state variables of the compartmental lag subsystems are hidden variables in the formulation with explicit lags.     

Resonant frequency of the chest-lung system by analysis of the respiratory flow curve

1. Resonant frequency of the chest-lung system in six tracheotomized, spontaneously breathing dogs was determined by analyzing the shape of the respiratory flow curve. 2. The resonant frequency was calculated from the periodic deviations from a sinusoidal flow pattern observed in the inspiratory phase of the breathing cycle. 3. Mean (+/- S.D.) resonant frequency was 6.1 +/- 0.9 Hz which was very close to the panting frequency (5.7 Hz) of the same dogs. 4. Resonant frequencies of the respiratory system in various species are compared.     

Application of the correlation integral to respiratory data of infants during REM sleep

 Non-linear time sequence analysis has been performed on infant sleep measurement data in order to obtain more information about the respiratory processes. As a first step, respiration data during REM sleep were analysed with methods from non-linear dynamics, especially, the correlation integral and the slope of its log-log plot, representing the correlation dimension. Before calculation of the correlation integral, a special kind of filtering has to be applied to the data. This filtering algorithm is a state space and singular value decomposition-based noise reduction method, and it is used to separate the noise and signal subspaces. The dynamics of a signal (in our case data from the respiratory process) and its degrees of freedom can be characterised by the correlation integral and by the correlation dimension, respectively. The main result of this study is that the highly irregular-looking breathing patterns during REM sleep could be described by a deterministic system, and finally the physiological significance of this finding is discussed. Received: 17 June 1994/Accepted in revised form: 18 November 1994     

Subset selection for linear mixed models

Linear mixed models (LMMs) are instrumental for regression analysis with structured dependence, such as grouped, clustered, or multilevel data. However, selection among the covariates—while accounting for this structured dependence—remains a challenge. We introduce a Bayesian decision analysis for subset selection with LMMs. Using a Mahalanobis loss function that incorporates the structured dependence, we derive optimal linear coefficients for (i) any given subset of variables and (ii) all subsets of variables that satisfy a cardinality constraint. Crucially, these estimates inherit shrinkage or regularization and uncertainty quantification from the underlying Bayesian model, and apply for any well-specified Bayesian LMM. More broadly, our decision analysis strategy deemphasizes the role of a single “best” subset, which is often unstable and limited in its information content, and instead favors a collection of near-optimal subsets. This collection is summarized by key member subsets and variable-specific importance metrics. Customized subset search and out-of-sample approximation algorithms are provided for more scalable computing. These tools are applied to simulated data and a longitudinal physical activity dataset, and demonstrate excellent prediction, estimation, and selection ability.     

Influence of gradual adaptation of cattle to Leucaena leucocephala leaf meal on biodegradation of mimosine and 3- hydroxy-4(1H)-pyridone (3,4 DHP) in rumen,their levels in blood,fate and influence of absorbed DHP on thyroid hormones and liver enzymes

Three rumen fistulated Karan Fries crossbred (Holstein X Tharparkar) calves were fed increasing dry matter (DM) levels of 25%, 50%, 75% and 100% through leucaena leaf meal (LLM) starting at week 1, 2, 3 and 4, respectively. The mimosine, 3,4 DHP and 2,3 DHP levels were determined in strained rumen liquor (SRL) and serum at 0, 1, 2, 4, 8, 12 and 24 h postfeeding on days 1, 8, 15, 22, 29 and 42. LLM was incubated for 24 h with SRL in vitro on days 0, 7, 14, 21, 28 and 41 to study mimosine and dihydroxypyridone (DHP) biodegradation. On day 43, 1–1.5 l of rumen liquor was transferred to another set of three unadapted calves which were fed 50% LLM after transfer of inoculum. DM intake was 1.78%, 2.13%, 2.27%, 1.66%, 1.54% and 1.35% of live weight during the 1st through 5th week, respectively. Both in vitro and in vivo studies showed extensive degradation of mimosine to 3,4 DHP and 2,3 DHP from first day of LLM feeding. The overall in vitro DHP degradation was nil, 28.6%, 43.3% and 40.1% on day upto 15, 21, 28 and 42 of LLM feeding. No mimosine was found in serum on any day of sampling. The 3,4 DHP detected (56.94±31.65 μg ml⁻¹ serum) one hour post feeding on day 1 exhibited a decline from day 22 onwards. The serum also contained 2,3 DHP on days 8, 15, 22, 42. The faecal and urinary excretion of 3,4 DHP and 2,3 DHP as percent of mimosine intake declined from first week (76.3±2.8) to 4th week (42.1±4.1). The feeding of LLM resulted in reduced level of T₃ and T₄ within a week of LLM feeding. The level of T₃ improved to normal by 6th week while that of T₄ remained low. The SGOT and SGPT activities were within normal range. The gradual adaptation to LLM feeding caused Karan Fries calves to acquire DHP degrading ability to nontoxic compounds and this ability was transferred through transfer of rumen liquor from such calves to other unadapted calves at as early as 9th day of LLM feeding. The results revealed the possibility of two types of microbes degrading mimosine and 3,4 DHP to 2,3 DHP. One type of 2,3 DHP degrading microbes may be inhibited in the presence of 3,4 DHP whereas the other type may be active.     

Non-linear, non-invasive method for seizure anticipation in focal epilepsy

In this paper we discuss an approach, using methods of non-linear time series analysis applied to scalp electrode recordings, which is able to distinguish between epochs temporally distant from and just prior to, the onset of a seizure in patients with temporal lobe epilepsy. The method involves a comparison of recordings taken from electrodes adjacent to and remote from the site of ictal onset. In particular, we define a non-linear quantity which we call 'marginal predictability'. This quantity is computed using data from remote and from adjacent electrodes. We find that the difference between the marginal predictabilities computed for the remote and adjacent electrodes decreases several tens of minutes prior to seizure onset, compared to its value interictally.     

WGCNA: an R package for weighted correlation network analysis

Background

Modelling the time-related behaviour of biological systems is essential for understanding their dynamic responses to perturbations. In metabolic profiling studies, the sampling rate and number of sampling points are often restricted due to experimental and biological constraints.

Results

A supervised multivariate modelling approach with the objective to model the time-related variation in the data for short and sparsely sampled time-series is described. A set of piecewise Orthogonal Projections to Latent Structures (OPLS) models are estimated, describing changes between successive time points. The individual OPLS models are linear, but the piecewise combination of several models accommodates modelling and prediction of changes which are non-linear with respect to the time course. We demonstrate the method on both simulated and metabolic profiling data, illustrating how time related changes are successfully modelled and predicted.

Conclusion

The proposed method is effective for modelling and prediction of short and multivariate time series data. A key advantage of the method is model transparency, allowing easy interpretation of time-related variation in the data. The method provides a competitive complement to commonly applied multivariate methods such as OPLS and Principal Component Analysis (PCA) for modelling and analysis of short time-series data.     

Modelling of the control of heart rate by breathing using a kernel method

The process of the breathing (input) to the heart rate (output) of man is considered for system identification by the input-output relationship, using a mathematical model expressed as integral equations. The integral equation is considered and fixed so that the identification method reduces to the determination of the values within the integral, called kernels, resulting in an integral equation whose input-output behaviour is nearly identical to that of the system. This paper uses an algorithm of kernel identification of the Volterra series which greatly reduces the computational burden and eliminates the restriction of using white Gaussian input as a test signal. A second-order model is the most appropriate for a good estimate of the system dynamics. The model contains the linear part (first-order kernel) and quadratic part (second-order kernel) in parallel, and so allows for the possibility of separation between the linear and non-linear elements of the process. The response of the linear term exhibits the oscillatory input and underdamped nature of the system. The application of breathing as input to the system produces an oscillatory term which may be attributed to the nature of sinus node of the heart being sensitive to the modulating signal the breathing wave. The negative-on diagonal seems to cause the dynamic asymmetry of the total response of the system which opposes the oscillatory nature of the first kernel related to the restraining force present in the respiratory heart rate system. The presence of the positive-off diagonal of the second-order kernel of respiratory control of heart rate is an indication of an escape-like phenomenon in the system.