Search

Bei der Uberwachung kritisch kranker Patienten mittels moderner Computer- und Messtechnologien fallen immense Datenmengen an. Die intelligente statistische Analyse dieser Daten in Echtzeit liefert wichtige Informationen uber den Gesundheitszustand und notwendige medizinische Masnahmen, um Leben zu retten und falsche Alarme zu vermeiden.

The purpose of this paper is to automatically detect local intensity changes in time series of grayscale images. For each pixel coordinate, a time series of grayscale values is extracted. An intensity change causes a jump in the level of the time series and affects several adjacent pixel coordinates at almost the same points in time. We use two-sample tests in moving windows to identify these jumps. The resulting candidate pixels are aggregated to segments using their estimated jump time and coordinates. As an application we consider data from the plasmon assisted microscopy of nanosize objects to identify specific particles in a sample fluid. Tests based on the one-sample Hodges-Lehmann estimator, the two-sample t-test or the two-sample Wilcoxon rank-sum test achieve high detection rates and a rather precise estimation of the change time.

International audience; A new fully automatic procedure for the construction of histograms is proposed. It consists of constructing both a regular and an irregular histogram and then choosing between the two. To choose the number of bins in the irregular histogram, two different penalties motivated by recent work in model selection are proposed. A description of the algorithm and a proper tuning of the penalties is given. Finally, different versions of the procedure are compared to other existing proposals for a wide range of densities and sample sizes. In the simulations, the squared Hellinger risk of the new procedure is always at most twice as large as the risk of the best of the other methods. The procedure is implemented in the R-Package histogram available from CRAN.

Introduction Monitoring of physiologic parameters in critically ill patients is currently performed by threshold alarm systems with high sensitivity but low specificity. As a consequence, a multitude of alarms are generated, leading to an impaired clinical value of these alarms due to reduced alertness of the intensive care unit (ICU) staff. To evaluate a new alarm procedure, we currently generate a database of physiologic data and clinical alarm annotations. Methods Data collection is taking place at a 12-bed medical ICU. Patients with monitoring of at least heart rate, invasive arterial blood pressure, and oxygen saturation are included in the study. Numerical physiologic data at 1-second intervals, monitor alarms, and alarm settings are extracted from the surveillance network. Bedside video recordings are performed with network surveillance cameras. Results Based on the extracted data and the video recordings, alarms are clinically annotated by an experienced physician. The alarms are categorized according to their technical validity and clinical relevance by a taxonomy system that can be broadly applicable. Preliminary results showed that only 17% of the alarms were classified as relevant, and 44% were technically false. Discussion The presented system for collecting real-time bedside monitoring data in conjunction with video-assisted annotations of clinically relevant events is the first allowing the assessment of 24-hour periods and reduces the bias usually created by bedside observers in comparable studies. It constitutes the basis for the development and evaluation of “smart” alarm algorithms, which may help to reduce the number of alarms at the ICU, thereby improving patient safety.

This paper compares the methods for variability extraction from a univariate time series in real time. The online scale estimation is achieved by applying a robust scale functional to a moving time window. Scale estimators based on the residuals of a preceding regression step are compared with regression-free and model-free techniques in a simulation study and in an application to a real time series. In the presence of level shifts or strong non-linear trends in the signal level, the model-free scale estimators perform especially well. However, the investigated regression-free and regression-based methods have higher breakdown points, they are applicable to data containing temporal correlations, and they are much more efficient.

A fast update algorithm for online calculation of the Q"n scale estimator is presented. This algorithm allows robust analysis of high-frequency time series in real time. It provides reliable estimates of a time-varying volatility even if many large outliers are present and it offers good efficiency in the case of clean Gaussian data.

INTRODUCTION Cardiovascular monitoring alarms are frequent in intensive care units (ICUs) and lead to noise levels often exceeding 80 dB. The aim of this study was to evaluate if there are relevant differences between ICUs with different subspecialties in the frequency and distribution of alarm signals, their occurrence during the day, the types of alarms and the underlying vital parameters. METHODS All alarm signals of the cardiovascular monitoring systems from randomly chosen patients at five different ICUs of the university hospital of Regensburg were evaluated. RESULTS No significant differences between the ICUs regarding the frequency of alarm signals and only slight differences in the time distribution could be recognized (p=0.02). The most frequent alarm signals were from threshold alarms (61%) followed by technical alarms. The majority of alarms generated were related to invasive arterial blood pressure measurement. CONCLUSIONS The frequency and distribution of ICU alarm signals seem to be comparable on different ICUs. Therefore, implementation of universal concepts for alarm reduction seems to be applicable regardless of the subspecialty of the ICU.

We propose a new regression-based filter for extracting signals online from multivariate high frequency time series. It separates relevant signals of several variables from noise and (multivariate) outliers. Unlike parallel univariate filters, the new procedure takes into account the local covariance structure between the single time series components. It is based on high-breakdown estimates, which makes it robust against (patches of) outliers in one or several of the components as well as against outliers with respect to the multivariate covariance structure. Moreover, the trade-off problem between bias and variance for the optimal choice of the window width is approached by choosing the size of the window adaptively, depending on the current data situation. Furthermore, we present an advanced algorithm of our filtering procedure that includes the replacement of missing observations in real time. Thus, the new procedure can be applied in online-monitoring practice. Applications to physiological time seri...

International audience; Even for a well-trained statistician the construction of a histogram for a given real-valued data set is a difficult problem. It is even more difficult to construct a fully automatic procedure which specifies the number and widths of the bins in a satisfactory manner for a wide range of data sets. In this paper we compare several histogram construction procedures by means of a simulation study. The study includes plug-in methods, cross-validation, penalized maximum likelihood and the taut string procedure. Their performance on different test beds is measured by their ability to identify the peaks of an underlying density as well as by Hellinger distance.

We consider data consisting of photon counts of diffracted x-ray radiation as a function of the angle of diffraction. The problem is to determine the positions, powers and shapes of the relevant peaks. An additional difficulty is that the power of the peaks is to be measured from a baseline which itself must be identified. Most methods of de-noising data of this kind do not explicitly take into account the modality of the final estimate. The residual-based procedure we propose uses the so-called taut string method, which minimizes the number of peaks subject to a tube constraint on the integrated data. The baseline is identified by combining the result of the taut string with an estimate of the first derivative of the baseline obtained using a weighted smoothing spline. Finally, each individual peak is expressed as the finite sum of kernels chosen from a parametric family., Comment: Published in at http://dx.doi.org/10.1214/08-AOAS181 the Annals of Applied Statistics (http://www.imstat.org/aoas/) by the Institute of Mathematical Statistics (http://www.imstat.org)

A definition of the ordering of a multivariate sample—based on the isobar surfaces—is used in order to obtain limit laws for the extreme values of a multidimensional sample.

Robustified rank tests, applying a robust scale estimator, are investigated for reliable and fast shift detection in time series. The tests show good power for sufficiently large shifts, low false detection rates for Gaussian noise and high robustness against outliers. Wilcoxon scores in combination with a robust and efficient scale estimator achieve good performance in many situations.

Current alarm systems in intensive care units create a very high rate of false positive alarms because most of them simply compare physiological measurements to fixed thresholds. An improvement can be expected when the actual measurements are replaced by smoothed estimates of the underlying signal. However, classical filtering procedures are not appropriate for signal extraction, as standard assumptions, such as stationarity, do no hold here: the time series measured often show long periods without change, but also upward or downward trends, sudden shifts and numerous large measurement artefacts. Alternative approaches are needed to extract the relevant information from the data, i.e., the underlying signal of the monitored variables and the relevant patterns of change, such as abrupt shifts and trends. This article reviews recent research on filter-based online signal extraction methods designed for application in intensive care.

Standard median filters preserve abrupt shifts (edges) and remove impulsive noise (outliers) from a constant signal but they deteriorate in trend periods. Finite impulse response median hybrid (FMH) filters are more flexible and also preserve shifts, but they are much more vulnerable to outliers. Application of robust regression, in particular of the repeated median, has been suggested for removing subsequent outliers from a signal with trends. A fast algorithm for updating the repeated median in linear time using quadratic space has been given by Bernholt and Fried (Inform. Process. Lett. 88 (2003) 111). Repeated median hybrid filters are constructed to combine the robustness of the repeated median with the edge preservation of FMH filters. Analytical properties of these filters are investigated and their performance is compared via simulations. An algorithm for updating the repeated median is presented which needs only linear space.

In intensive care, time series of vital parameters have to be analysed online, i.e. without any time delay, since there may be serious consequences for the patient otherwise. Such time series show trends, slope changes and sudden level shifts, and they are overlaid by strong noise and many measurement artefacts. The development of update algorithms and the resulting increase in computational speed allows to apply robust regression techniques to moving time windows for online signal extraction. By simulations and applications we compare the performance of least median of squares, least trimmed squares, repeated median and deepest regression for online signal extraction.

Data from the automatic monitoring of intensive care patients exhibits trends, outliers, and level changes as well as periods of relative constancy. All this is overlaid with a high level of noise and there are dependencies between the different items measured. Current monitoring systems tend to deliver too many false warnings which reduces their acceptability by medical staff. The challenge is to develop a method which allows a fast and reliable denoising of the data and which can separate artefacts from clinical relevant structural changes in the patients condition (Estadistica 53 (2001) 259). A simple median filter works well as long as there is no substantial trend in the data but improvements may be possible by approximating the data by a local linear trend. As a first step in this programme, the paper examines the relative merits of the L1 regression, the repeated median (Biometrika 68 (1982) 242) and the least median of squares (Bull. Internat. Statist. Inst. 46 (1975) 375; J. Amer. Statist. Assoc. 79 (1984) 871). The question of dependency between different items is a topic for future research.

Exponential and normal distributions will be characterized by the independence of linear forms and in order statistics X 1, n ≤ … ≤ X n, n . As a by-product, we show that no L-statistic with a i ≥ 0 and can be a complete and sufficient maximum likelihood estimator of a location parameter except from the arithmetic mean X¯ in the normal and X 1, n (X n, n ) in the exponential (reflected exponential) case.

The detection of patterns in monitoring data of vital signs is of great importance for adequate bedside decision support in critical care. Currently used alarm systems, which are based on fixed thresholds and independency assumptions, are not satisfactory in clinical practice. Time series techniques such as AR-models consider autocorrelations within the series, which can be used for pattern recognition in the data. For practical applications in intensive care the data analysis has to be automated. An important issue is the suitable choice of the model order which is difficult to accomplish online. In a comparative case-study we analyzed 34564 univariate time series of hemodynamic variables in critically ill patients by autoregressive models of different orders and compared the results of pattern detection. AR(2)-models seem to be most suitable for the detection of clinically relevant patterns, thus affirming that treating the data as independent leads to false alarms. Moreover, using AR(2)-models affords only short estimation periods. These findings for pattern detection in intensive care data are of medical importance as they justify a preselection of a model order, easing further automated statistical online analysis.

Sliced Inverse Regression (SIR) is a promising technique for the purpose of dimension reduction. Several properties of this method have been examined already, but little attention has been paid to robustness aspects. In this article, we focus on the sensitivity of SIR to outliers and show in what sense and how severely SIR can be influenced by outliers in the data.

The aim of detecting outliers in a multivariate sample can be pursued in different ways. We investigate here the performance of several simultaneous multivariate outlier identification rules based on robust estimators of location and scale. It has been shown that the use of estimators with high finite-sample breakdown point in such procedures yields a good behaviour with respect to the prevention of breakdown by the masking effect (Becker, Gather, 1999. J. Amer. Statist. Assoc. 94, 947–955). In this article, we investigate by simulation, at which distance from the centre of an underlying model distribution outliers can be placed until certain simultaneous identification rules will detect them as outliers. We consider identification procedures based on the minimum volume ellipsoid, the minimum covariance determinant, and S-estimators.

We consider the standardized median as an estimator of scale for exponential samples which is most B-robust in the sense of Hampel et al. (1986). This estimator is compared with two other estimators which were proposed to Rousseeuw and Croux (1993) but for a Gaussian model. All three estimators have the same breakdown point, but their bias curves are different. It is shown that under a gross error model the explosion bias curve of the most B-robust estimator performs better than the bias curves of the other estimators. But this estimator is worse than the two estimators proposed by Rousseeuw and Croux (1993) if the implosion bias curve is considered.

Several jackknife estimators of a relative risk in a single 2×2 contingency table and of a common relative risk in a 2×2×K contingency table are presented. The estimators are based on the maximum likelihood estimator in a single table and on an estimator proposed by Tarone (1981) for stratified samples, respectively. For the stratified case, a sampling scheme is assumed where the number of observations within each table tends to infinity but the number of tables remains fixed. The asymptotic properties of the above estimators are derived. Especially, we present two general results which under certain regularity conditions yield consistency and asymptotic normality of every jackknife estimator of a bunch of functions of binomial probabilities.

In order to describe or generate so-called outliers in univariate statistical data, contamination models are often used. These models assume that k out of n independent random variables are shifted or multiplicated by some constant, whereas the other observations still come i.i.d. from some common target distribution. Of course, these contaminants do not necessarily stick out as the extremes in the sample. Moreover, it is the amount and magnitude of ‘contamination” which determines the number of obvious outliers. Using the concept of Davies and Gather (1993) to formalize the outlier notion we quantify the amount of contamination needed to produce a prespecified expected number of ‘genuine’ outliers. In particular, we demonstrate that for sample of moderate size from a normal target distribution a rather large shift of the contaminants is necessary to yield a certain expected number of outliers. Such an insight is of interest when designing simulation studies where outliers shoulod occur as well as in theo...