Patients and ethical considerations
The study was carried out according to the Declaration of Helsinki and the guidelines of the International Con- ference for Harmonization for Good Clinical Practice, it was approved by the local Ethics Committee of the Uni- versity Hospital Essen (Institutional Review Board). In a prospective monocenter study (11/2006–12/2010), we recruited 108 consecutive ALF patients (64% females / 36% males), who met the criteria defined by the “Acute Liver Failure Study Group Germany” . In brief, ALF was diagnosed by significant liver dysfunction with pathologically increased laboratory parameters (bilirubin, AST, ALT, AP, γ-GT) and an international normalized ratio (INR) of >1.5 with the concomitant presence of any degree of encephalopathy. Reference values for normal ranges are presented in Table 1. A pathological increase was defined as any value above these ranges. Other causes of liver dysfunction were excluded, such as acute- on-chronic liver failure or pre-existing cirrhosis. All patients had presented within four weeks of disease on- set without pre-existing liver disease. Upon admission, clinical data were collected. Outcome (spontaneous re- covery, SR; non-spontaneous recovery, NSR: comprising transplantation or death) was defined by the status after 4 weeks post admission.
Hospital Essen, the appropriate institutional review board. Informed written consent was given by probands and postoperative patients. Informed consent of patients with sepsis was waived by the ethics committee, but written informed consent for the use of data was acquired from the surviving patients after recovery from the disease. Over a period of two years, 161 patients admitted to an intensive care unit (ICU) of the Univer- sity Hospital of Essen were considered eligible for the study if they fulfilled the criteria for severe sepsis (sepsis group) . The second cohort comprised patients admitted to the ICU after surgery, who did not meet the criteria for sepsis (postoperative group, n = 136). Two patients in this group died in the postoperative course. Moreover, free hemoglobin concentration was deter- mined in 9 healthy probands. A detailed characterization of patients with sepsis and postoperative controls is given in Table 1. Plasma obtained from EDTA blood, drawn for standard laboratory determinations and cen- trifuged within 2 hours (15 minutes, 2,500 g), was used for determination of free hemoglobin on day 1 of diag- nosis and surgery, respectively. Plasma samples were stored frozen at -70 ° C. Blood was drawn via an indwel- ling arterial catheter (postoperative and septic patients) or via venous puncture (probands). Procalcitonin con- centration, the simplified acute physiology score II (SAPS II), and the sequential organ failure assessment score (SOFA score) were determined on day 1 [11,12]. The amount of packed red blood cell transfusions within 3 days and 4 weeks before surgery or the diagno- sis of sepsis, and 30-day survival were also determined.
decrease in extracellular fluid osmolality is associated with an increase in brain swelling, factors that modulate fluid osmolality were taken into account as prognostic markers [14,15]. Patients with serum sodium between 145 and 150 mmol/l are known to have fewer episodes of intracranial hypertension and consequently a higher risk to develop brain edema. Thus, hyponatremia might worsen the prog- nosis in ALF . As hyponatremia is associated with poor prognosis in cirrhosis, inclusion of serum sodium (Na) into the MELD was found to improve its predictive value in chronic liver diseases [13,17]. Two sodium containing MELD modifications, “UKELD” and “MELD-Na”, were proposed to enhance its prognostic ability in chronic liver failure [18,19]. A potential predictive value of these modifi- cations in ALF has not been evaluated yet.
and biological parameters (which were also used for calculation of the MIPI scores), using a machine-learning algorithm, showed the same trend: best results were achieved when all parameters were integrated into a single model. We used an MLP neural network for outcomeprediction, which is a universal function approximator with the ability to model any type of regression or classification problem [ 37 ]. While MLP networks are well established in the machine-learning community as powerful pre- diction algorithms [ 38 ], it is possible that even more advanced, deep machine-learning techniques such as convolutional neural networks (CNN), with their larger numbers of hidden layers and their interconnection between neurons within the same layer, may have performed even better. However, CNNs are mainly intended for use with large datasets (“big data”). In a patient population such as our own, however, their complexity would have increased the probability of “overfitting”, i.e. loss of gener- alizability of the model, in our case for PFS prediction [ 39 ].
empirical criterion from Choi/Chang has lead to better results . A serious limitation of the modelling approach in CODAC using 2-D shell elements is it’s inability to properly model de- laminations, which split the laminate into two or more sub-laminates. This failure mode leads to a reduction in bending stiffness and strongly influences the transverse shear distribution. A proper model of a delaminated structure would require e.g. a model of stacked shells and a method to deal with the contact problem between the different sub-laminates. However, it will be shown here that it is possible to quickly achieve acceptable results with the presented simplified methodology.
Some studies have examined the issue of what should be the knowledge and skills components of today’s accounting education programs that can satisfy the demands for training future accountants at university level (Mihret & Bobe, 2014; Kidane, 2012; ROSC, 2007) . In addition, the challenges of the existing TVET system were studied in general terms to some extent (Dadi, 2014; Solomon, 2011). But accounting education and training at TVET level in particular was relatively ignored. Hence, there is increasing recognition by all stakeholders such as teachers, students, employers, governments and society the need and utility of conducting research within TVET colleges in order to position such institutions for the challenges of the 21 st century. The main objective of this study is to examine the limitations of the existing TVET in Accounting Curriculum. Specifically, the study is aimed to investigate the existing TVET in accounting curriculum in terms of relevance to local needs, to examine the existing TVET in accounting curriculum in terms of its content as compared to international standard and to identify competences that were not usually covered in the curriculum and the reasons behind.
Lebensminuten vorhanden sind. Unter Berücksichtigung genannter Limitationen einzelner Arbeiten zeigen die Resultate von Geissbühler et al. (2004), Schröcksnadel et al. (2003) und Zanetti-Dällenbach et al. (2006) ein signifikant besseres fetales Outcome nach einer Wassergeburt, im Vergleich zu Kinder, welche ausserhalb des Wassers zur Welt kommen. Die Resultate von Cluett et al. (2009) zeigen kein besseres, aber auch kein schlechteres fetales Outcome nach Wassergeburt. Sie zeigen jedoch ein gleich gutes fetales Outcome, ob Wasser- oder Landgeburt, sofern Richtlinien eingehalten werden. Werden solche missachtet, können Komplikationen eintreten, wie dies die gesamte analysierte Literatur zu erkennen gibt.
Die Hemikraniektomie bei maligner Hirnschwellung unterschiedlicher Genese ist in der Lage, die Sterblichkeitsrate dieser Patienten signifikant zu senken. Die Chance, dadurch aber auch ein zumindest teilselbstständiges Leben wiederzuerlangen ist für die verschiedenen Ursachen sehr unterschiedlich. Insgesamt sahen wir zwar bei 40 % der überlebenden Patienten ein gutes outcome. Die beste Prognose hatten jedoch Patienten mit einem Schädelhirntrauma mit 71 % gutem outcome, wohingegen keiner unserer Patienten mit einem malignen Mediainfarkt in diese Gruppe einzuordnen war. Zusätzlich gilt, dass je älter der Patient ist, umso schlechter sein outcome zu erwarten sein wird. Wird die Entlastungskraniektomie zu spät durchgeführt, d.h. ist bereits eine klinische Verschlechterung eingetreten, sinkt die Chance auf ein gutes outcome dramatisch.
Auch für Krankenhäuser insgesamt wurde ein Volumeneffekt nachgewiesen. Bezüglich Ösophagus und Pankreas wurde beispielsweise gefunden, dass Patienten in Kliniken mit hoher Fallzahl eine geringere Mortalität aufweisen. Auch die Langzeitprognose dieser Patienten ist besser als die der Patienten von Häusern mit geringerer Fallzahl. Tumorchirurgie des Ösophagus und Pankreas sollte in Kliniken mit hoher Fallzahl ausgeführt werden (Hölscher 2004). Stockton et al. verglichen retrospektiv das Outcome spezialisierter Tumorzentren mit dem Outcome von Kreiskrankenhäusern. Es wurden die Daten von 14 527 Patienten ausgewertet. Die Tumorzentren zeigten auch hier signifikant bessere Ergebnisse (Stockton 2000). Selbst in high-volume Krankenhäusern können die Patienten allerdings ihre Überlebenschance verbessern, wenn sie Chirurgen wählen, die diese Operation häufig durchführen (Birkmeyer, 2003). Verständlicherweise muss ein hohes Volumen eines Chirurgen so aussehen, dass dieser Chirurg sich auf einige wenige bestimmte Operationen spezialisiert, und diese sehr häufig durchführt. Es kann nicht sein, dass ein Chirurg versucht, sein Volumen einer bestimmten Operation zu erhöhen indem er zusätzlich zu seiner bisherigen Bandbreite einfach mehr Patienten, welche dieser bestimmten Operation bedürfen, behandelt. Denn dann hätte der Chirurg für den einzelnen Patienten noch weniger Zeit und Aufmerksamkeit, was wiederum zu einem schlechteren Outcome führen würde.
Transplantatüberleben haben, an Empfänger vergeben, die eine prognostiziert vergleichbare verbleibende Lebenserwartung nach einer Transplantation haben. Um diesen Voraussetzungen gerecht zu werden, wurde neben dem bereits existierenden „Eurotransplant Kidney Allocation System“ (ETKAS) 1999 das „Eurotransplant Senior Programm“ (ESP) ins Leben gerufen, an dem Patienten, die mindestens 65 Jahre alt sind, freiwillig teilnehmen können. Hier werden Nieren von Spendern, die ≥65 Jahre alt sind, an Empfänger, die ≥65 Jahre alt sind, ohne Berücksichtigung des HLA-Matches lokal vergeben (51). Das Eurotransplant Senior Programm zielt dabei darauf ab, eine kurze kalte Ischämiezeit, eine verminderte Wartezeit und ein gutes funktionelles Match zu erreichen. Eine Analyse aus dem Jahr 2008 von Frei et al. konnte zeigen, dass diese Ziele erreicht wurden. Es fand sich ein akzeptables Transplantatüberleben mit einem hohen Anteil von Patienten, bei denen das Transplantat bis zum Ende des Lebens funktionierte (14). Es gibt jedoch weniger Daten, die detailliert das immunologische Langzeit-Outcome bei diesen älteren Patienten untersuchen. Sehr viele Studien zeigen übereinstimmend, dass HLA-Mismatches zur Entwicklung von donorspezifischen Antikörpern, Rejektionen und schlechterem Transplantatüberleben führen können. Das legt nahe, dass insbesondere ältere Transplantatempfänger, bei denen die Allokation im Rahmen des Eurotransplant Senior Programms ohne Berücksichtigung des HLA-Matchings vorgenommen wurde, ein besonders hohes Risiko für immunologische Komplikationen haben. Dabei sind gerade diese Patienten anfällig für die Nebenwirkungen von intensiven immunosuppressiven Induktions-, Erhaltungs- oder Rejektionstherapien (50, 52, 53).
8. Donovan JL, Hamdy FC, Lane JA, Mason M, Metcalfe C, Walsh E, Blazeby JM, Peters TJ, Holding P, Bonnington S et al: Patient-Reported Outcomes after Monitoring, Surgery, or Radiotherapy for Prostate Cancer. New England Journal of Medicine 2016, 375(15):1425- 1437.
gation models in order to determine the best parameters for a non-premixed ethylene/air flame [ 26 ]. Three different kinds of inception models (acetylene in- ception route, the PAH inception route, and the naphthalene inception route), three types of surface area functions for growth, three different coagulation constants and finally three expressions of soot oxidation were investigated. Zucca et al. proposed another model combination for the same type of flames, again validated against experimental results [ 30 ]. Being more recent, the model formulation presented in [ 30 ] has been retained as initial solution for the op- timization process described in what follows. Even if such studies have led to a significant improvement in understanding and predicting soot formation, available models have yet to yield satisfactory comparisons against all avail- able experimental data [ 31 , 32 ]. There is thus a clear need for more accurate prediction models, applicable to a wide range of flow and operating conditions. First computational studies in our group concerning predictions of soot vol- ume fraction [ 33 ] have demonstrated that the nucleation and oxidation models play a first-order role for typical flame configurations, followed by molecular growth and finally aggregation processes. Therefore, nucleation and oxidation parameters are first retained in the following optimization.
As the FETCPM change point detector can operate on cumulated data, such as the sufficient statistics R n and S n (Equation 19 , p. 147 ), as well as directly on the time series representation, it is easily applicable to both the compact standard CertainTrust formulation, as well as sam- ples in 0 − 1 random vector form as used in the previous sections of this thesis. The FETCPM detector, as described above, monitors for changes in a binomial proportion. Thus, in the case of trust assessment in the binomial case, the application of FETCPM is straightforward; one simply monitors the sum of failures for increases in the parame- ter p of the Bernoulli distributed process generating the observations, and the sum of successes for decreases in p, as well as a window of ad- equate size w. Once a change is detected by the change detector, the observations made at time points before the signalled change point are simply discarded. This is justified, because those observations are assumed to be generated by a different Bernoulli process than the one currently active. As a consequence, change detection implements abrupt forgetting conditional on the thresholded statistic of the change detector. Initialising the FETCPM change detector requires observing a minimum number of data points that are assumed to be indepen- dent and identically distributed (iid); the R-implementation of FETCPM in [ 177 ] recommends 20 observations.
To prevent LC accidents, ADASs have been developed to predict forthcoming driver LC behaviors while driving. For example, LC assistance system could assess the risk levels of maneuvering a LC under the current driving situation. If the driver intends to make LC with high risk, an alarm will be delivered to avoid the potential accidents. This function is depicted in Figure 2.1. Comprehensively understanding of driver LC behavior is very important but nontrivial. Fortunately, carefully partitioning the LC procedure into small segments could make this goal achievable. From the perception-action level, a complete LC task can be roughly divided into three stages: forming intent, preparing actions , and executing actions. The driver first forms the LC intent according to his/her traveling plan and the current driving situation, and with such intent, prepares for taking LC actions by longitudinal adjustment (e.g., waiting, accelerating, or decelerating), and then executes a series of LC actions such as lateral controls as long as the driving situation is acceptable (Windridge, Shaukat, and Hollnagel, 2013). To ensure safe driving, predicting driver LC behavior as early as possible can leave enough time to prevent improper LC behaviors. In this chapter we will give an overview of the work regarding to the predictionof driver LC behavior from basic concept to methodology which will lay the foundation of the entire work.
In this paper, DLR’s empirical prediction model for flap tip noise is presented and discussed in details. The prediction scheme is based on a comprehensive acoustic and aerodynamic database acquired in the Acoustic Wind Tunnel Braunschweig. It was verified, through successful scaling of the measured noise spectra, that the cross-flow velocity at the flap tip is an important parameter characterizing the flow mechanism(s) responsible for the noise production. This finding led to the definition of a universal flap tip noise spectral shape in terms of a linear least-squares fit of the corresponding measurement data. Using a similar approach, a model for the flap tip noise direc- tivity was formulated. The prediction model was compared against full-scale fly-over measurement data (B747-400 and A319) and an acceptable agreement of the overall predictions was found. A slight underprediction of the noise levels at high frequen- cies suggests that additional airframe noise sources might be needed in the complete aircraft noise prediction scheme to get a better agreement between measured and pre- dicted noise levels. It is also found that, for large flap deflection angles, flap tip noise dominates the high frequency part of the predicted complete aircraft high-lift noise spectra. Knowledge of the flap tip noise peak frequency and high-frequency decay are therefore sufficient to account for this noise source in the total aircraft noise prediction. Finally, the limitations of the prediction scheme are discussed and research needs are identified.
When an accurate instrument is unavailable or prohibitively expensive but cer- tain data related to the cause of misclassification are available, one can develop an identifiable model in an attempt to correct for the misclassification bias in the esti- mators and predictors. For the single proportion problem using misclassified data with no training data, Gaba and Winkler (1992) and Viana et al. (1993) developed Bayesian approaches with highly informative priors. Bayesian inferences with in- formative priors were also developed for two-sample problems for two proportions. For example, see Evans et al. (1996) for risk difference (the difference of two pro- portions) and Gustafson et al. (2001) for odds ratios. Lahiri and Larsen (2005) used a mixture model to correct for the bias of the ordinary least square estimators of regression coefficients due to imperfect linkages.
183 investigation on the predictionof monomer retention in size exclusion chromatography (SEC). We coupled SEC with multi angle laser light scattering (MALS) which is useful for studying and characterizing aggregation due to its high sensitivity towards high molecular weight species. 18,19 SEC-MALS results provide an absolute means for qualitative and quantitative analysis of protein aggregation kinetics. 20–23 For the predictionof long term stability from accelerated stability studies we propose the use of artificial neural networks (ANNs). ANNs have been applied in the pharmaceutical research for many different purposes. 24–33 However, a significant limitation of ANNs is their interpretability. Algorithms, such as TREPANs, have been used to extract a comprehensible concept description from a trained network to solve classification problems. 34 For certain regression problems another approach is to use surrogate models. 35 Therefore, we investigated several machine learning algorithms (e.g. Support vector machines) as surrogate models to understand and build trust in the ANNs decision making process. Amongst the one investigated we selected “white box” algorithms (i.e. human friendly), namely linear regression and a decision trees, as no significant benefit was observed in applying more complex models. Fourteen therapeutically relevant proteins were investigated in twenty-four pharmaceutically relevant formulation conditions. Our hypothesis was that the output of suitably designed short term accelerated stability assays could be used to generate formulation dependent protein stability “finger-prints”. These features are used as input for the ANNs to predict the real-time stability of proteins. Furthermore, we developed multiple algorithms which included other additional high-throughput biophysical parameter typically used to investigate protein stability, e.g. the temperature of unfolding (T m ). We investigate ANN architectures
The ship’s X-band radars provide the surface elevation for domains of several square kilometres including information on directionality. Theoretically, the detected wave field can be available after every revolution of the radar antenna (<2 s), even though time is needed for the inversion of the radar clutter (due to the presence of the waves) to the wave field. In contrast to single point measurements, where a sufficient measuring time is needed for an accurate and sufficiently long prediction, this time lag is only limited by the numerical efficiency of the used algorithm. Unfortunately, the principle of measurements of X-band radars also yield a main drawback as the radar cannot detect wave troughs behind (steep) wave crests. Thus, the accuracy is dependent on the wave inversion algorithm. However, the advantages of using a ship’s X-band radar (plug-and-play, very fast wave registration over a large space domain, directionality of the wave field, and large prediction horizon) seems to outweigh the disadvantages as two commercial prediction systems are available from the companies Applied Physical Sciences Corporation (FutureWaves TM [ 5 ]) and Next Ocean TM [ 6 ]. Even this paper focusses on wave prediction exclusively, the presented research assumes surface elevation snapshots (taken continuously by a ship board radar at great distance ahead the operational area) as input for the deterministic wave prediction.
Upper extremity amputation can be related to accidental trauma, disease, defect or the progress of war. Although persons have suered arm amputation for centuries, it is only during the past 500 years that substitutes have been fabricated in an attempt to provide a substitute for the amputated part of the arm. Much of the early writing that exists in reference to amputation, is related to war injuries. One of the rst examples of an early articial hand is a mailed st that was made for Goetz von Berlichingen in the middle ages. This prosthesis was equipped with jointed ngers that could passively grips his sword like a vice. Even with the advances made in the electrical control of prosthetic part, man has not build an excellent substitute for a missing hand, wrist, elbow or shoulder (Meier and Atkins, 2004). In the context of non-invasive interfaces for controlling mechanical hands, a concrete possibility arises from forearm surface electromyography (EMG), a technique by which muscle activation potentials are gathered by electrodes placed on the patient's forearm. These potentials can be used to track which muscles the patient is activating, and with what force (Castellini and van der Smagt, 2009). The use of myoelectric signals for control seems to have been suggested rst by Reiter, in Germany, around 1948 (Wirta, et al., 1977).
Nonlinear and time-linearized gust load analysis based on RANS solutions are conducted cov- ering a big part of the gust parameter space in order to quantify nonlinear effects in the gust load prediction. The nonlinear loads of the wing bending moment display a much attenuated increase with increasing gust amplitudes in contrast to the predicted the time-linearized loads. Moreover, the maximum magnitude of the minimum wing bending moment is not found for the greatest gust length as it is predicted by the time-linearized analysis. The analysis of the unsteady nonlinear surface pressure reveals a distinct shock motion due to the gust encounter. The shock’s strength and its range of motion increases with increasing gust amplitude. Addi- tionally, the shock’s range of motion is significantly increased with increasing gust length. The time-linearized approach agrees very well with nonlinear loads for very small gust amplitudes but it clearly overpredicts the wing bending moment with increasing gust amplitude. Further- more, the unsteady shock characteristcs of the time-linearized surface pressure is found to only match the cases for very small gust amplitudes. For greater gust amplitudes the time-linearized range of shock motion is limited to the small gust amplitude case producing unphysical peaks in the surface pressure distribution. This leads to overpredicted as well as underpredicted wing torsional moments and thus, no general conclusion over the possible conservatism introduced by the time-linearized loads can be reached.