To pretrain or not? A systematic analysis of the benefits of pretraining in diabetic retinopathy

There is an increasing number of medical use cases where classification algorithms based on deep neural networks reach performance levels that are competitive with human medical experts. To alleviate the challenges of small dataset sizes, these systems often rely on pretraining. In this work, we aim to assess the broader implications of these approaches in order to better understand what type of pretraining works reliably (with respect to performance, robustness, learned representation etc.) in practice and what type of pretraining dataset is best suited to achieve good performance in small target dataset size scenarios. Considering diabetic retinopathy grading as an exemplary use case, we compare the impact of different training procedures including recently established self-supervised pretraining methods based on contrastive learning. To this end, we investigate different aspects such as quantitative performance, statistics of the learned feature representations, interpretability and robustness to image distortions. Our results indicate that models initialized from ImageNet pretraining report a significant increase in performance, generalization and robustness to image distortions. In particular, self-supervised models show further benefits to supervised models. Self-supervised models with initialization from ImageNet pretraining not only report higher performance, they also reduce overfitting to large lesions along with improvements in taking into account minute lesions indicative of the progression of the disease. Understanding the effects of pretraining in a broader sense that goes beyond simple performance comparisons is of crucial importance for the broader medical imaging community beyond the use case considered in this work.

Variation in CFHR3 determines susceptibility to meningococcal disease by controlling factor H concentrations

Neisseria meningitidis protects itself from complement-mediated killing by binding complement factor H (FH). Previous studies associated susceptibility to meningococcal disease (MD) with variation in CFH, but the causal variants and underlying mechanism remained unknown. Here we attempted to define the association more accurately by sequencing the CFH-CFHR locus and imputing missing genotypes in previously obtained GWAS datasets of MD-affected individuals of European ancestry and matched controls. We identified a CFHR3 SNP that provides protection from MD (rs75703017, p value = 1.1 × 10^-16) by decreasing the concentration of FH in the blood (p value = 1.4 × 10^-11). We subsequently used dual-luciferase studies and CRISPR gene editing to establish that deletion of rs75703017 increased FH expression in hepatocyte by preventing promotor inhibition. Our data suggest that reduced concentrations of FH in the blood confer protection from MD; with reduced access to FH, N. meningitidis is less able to shield itself from complement-mediated killing.

The descending pain modulation system predicts short term efficacy of multimodal pain therapy - an observational prospective cohort study

OBJECTIVES

Treating chronic pain patients with multimodal pain therapy (MMPT) alters perception, awareness, and processing of pain at multiple therapeutic levels. Several clinical observations suggest that the effects of therapy may go beyond the possible sum of each level of therapy and may be due to a central descending inhibitory effect measurable by conditioned pain modulation (CPM). Thus, we investigated whether CPM is able to identify a group of patients that benefit particularly from MMPT.

METHODS

This was an observational prospective cohort study. Patients were hospitalized on a special pain medicine ward with specially trained staff for 10 days. The patients were questioned and had investigations before and shortly after MMPT and were followed-up on 3 months post discharge. Before and after treatment, subjects were investigated via CPM and quantitative sensory testing (QST) as well as completing questionnaires. The study was registered in the German Clinical Trials Register (DRKS00006850).

RESULTS

During the study period of 24 months, 224 chronic pain patients were recruited. 51 percent of patients completed the study period. There was an improvement in overall groups regarding all domains assessed, lasting beyond the end of the intervention. Patients with a sufficient CPM effect, defined as a reduction in pain during the conditioning stimulus, at baseline did show a more pronounced reduction in mean pain ratings than those without. This was not the case 3 months after therapy. Furthermore, sufficient CPM was identified as a predictor for pain reduction using a linear regression model.

CONCLUSION

In conclusion, this study shows that while a heterogeneous group of patients with chronic pain disorders does sustainably benefit from MMPT in general, patients with a sufficient CPM effect do show a more pronounced decrease in pain ratings directly after therapy in comparison to those without.

Toward Scalable and Unified Example-Based Explanation and Outlier Detection

When neural networks are employed for high-stakes decision-making, it is desirable that they provide explanations for their prediction in order for us to understand the features that have contributed to the decision. At the same time, it is important to flag potential outliers for in-depth verification by domain experts. In this work we propose to unify two differing aspects of explainability with outlier detection. We argue for a broader adoption of prototype-based student networks capable of providing an example-based explanation for their prediction and at the same time identify regions of similarity between the predicted sample and the examples. The examples are real prototypical cases sampled from the training set via a novel iterative prototype replacement algorithm. Furthermore, we propose to use the prototype similarity scores for identifying outliers. We compare performance in terms of the classification, explanation quality and outlier detection of our proposed network with baselines. We show that our prototype-based networks extending beyond similarity kernels deliver meaningful explanations and promising outlier detection results without compromising classification accuracy.

Psychological burden, stressors and resources of social workers working with women and children who suffered extreme violence by the 'Islamic state': A mixed method study

OBJECTIVE

This explorative study aimed to determine the extent of psychological burden in social workers working with traumatized refugees. In addition, distressing and helpful factors determining the psychosocial burden were to be identified and described.

METHODS

Cross-sectional, mixed method design using quantitative and qualitative methods. The quantitative part included the Perceived Stress Questionnaire (PSQ) and items to assess specific factors of the working-context. The qualitative part is based on 5 focus groupdiscussions and 16 individual interviews. Evaluation was carried out using qualitative content analysis (QCA) including cross-analysis along the subscales of the PSQ to organise the qualitative material.

RESULTS

N = 54 social workers completed the questionnaire. High scores were found for all subscales of the PSQ. The distressing factor rated the highest was need of interpreters to communicate (M = 5.1, SD = 1.71), the helpful factor rated the highest was communication skills (M = 6.35, SD = 0.73). In the QCA, aspects of distressing and helpful factors were identified and further detailed.

CONCLUSION

According to the here presented study results, the psychological burdens of social workers working with refugees seem to be high. The impact of distressing factors such as working with interpreters and exposure to trauma content or PTSD symptoms might be reduced by offering specific education and supervision. The individual extent of psychological burden should be considered and (re-)evaluated on a regular basis as secondary prevention. Helpful factors like self-care, teamwork, networking and cooperation are evident and should be supported by implementing professional and psychological support.

Resolving challenges in deep learning-based analyses of histopathological images using explanation methods

Deep learning has recently gained popularity in digital pathology due to its high prediction quality. However, the medical domain requires explanation and insight for a better understanding beyond standard quantitative performance evaluation. Recently, many explanation methods have emerged. This work shows how heatmaps generated by these explanation methods allow to resolve common challenges encountered in deep learning-based digital histopathology analyses. We elaborate on biases which are typically inherent in histopathological image data. In the binary classification task of tumour tissue discrimination in publicly available haematoxylin-eosin-stained images of various tumour entities, we investigate three types of biases: (1) biases which affect the entire dataset, (2) biases which are by chance correlated with class labels and (3) sampling biases. While standard analyses focus on patch-level evaluation, we advocate pixel-wise heatmaps, which offer a more precise and versatile diagnostic instrument. This insight is shown to not only be helpful to detect but also to remove the effects of common hidden biases, which improves generalisation within and across datasets. For example, we could see a trend of improved area under the receiver operating characteristic (ROC) curve by 5% when reducing a labelling bias. Explanation techniques are thus demonstrated to be a helpful and highly relevant tool for the development and the deployment phases within the life cycle of real-world applications in digital pathology.

Memory snapshot dataset of a compromised host with malware using obfuscation evasion techniques

This article presents a dataset for studying the detection of obfuscated malware in volatile computer memory. Several obfuscated reverse remote shells were generated using Metasploit-Framework, Hyperion, and PEScrambler tools. After compromising the host, Memory snapshots of a Windows 10 virtual machine were acquired using the open-source Rekall's WinPmem acquisition tool. The dataset is complemented by memory snapshots of uncompromised virtual machines. The data includes a reference for all running processes as well as a mapping for the designated malware running inside the memory. The datasets are available in the article, for advancing research towards the detection of obfuscated malware from volatile computer memory during a forensic analysis.

[Pharmacotherapy of chronic neuropathic pain]

Chronic neuropathic pain has a prevalence of 6.9-10% in the general population. The current recommendations for treatment are presented based on a literature search. Neuropathic pain requires the use of co-analgesic, antidepressant, anticonvulsant drugs and topical agents because non-opioid analgesic drugs are usually ineffective. The use of meta-analyses tricyclic antidepressants, selective serotonin-norephinephrine reuptake inhibitors, and calcium channel anticonvulsants are recommended as the drugs of first choice. Under certain conditions chronic neuropathic pain can be treated with opioids. Topical therapeutics are only used to treat peripheral neuropathic pain. At present the use of drugs is independent of the etiology of the pain. Comorbidities, concomitant medication, potential side effects and patients' age have to be considered in treatment planning.

Identification of regulatory variants associated with genetic susceptibility to meningococcal disease

Non-coding genetic variants play an important role in driving susceptibility to complex diseases but their characterization remains challenging. Here, we employed a novel approach to interrogate the genetic risk of such polymorphisms in a more systematic way by targeting specific regulatory regions relevant for the phenotype studied. We applied this method to meningococcal disease susceptibility, using the DNA binding pattern of RELA – a NF-kB subunit, master regulator of the response to infection – under bacterial stimuli in nasopharyngeal epithelial cells. We designed a custom panel to cover these RELA binding sites and used it for targeted sequencing in cases and controls. Variant calling and association analysis were performed followed by validation of candidate polymorphisms by genotyping in three independent cohorts. We identified two new polymorphisms, rs4823231 and rs11913168, showing signs of association with meningococcal disease susceptibility. In addition, using our genomic data as well as publicly available resources, we found evidences for these SNPs to have potential regulatory effects on ATXN10 and LIF genes respectively. The variants and related candidate genes are relevant for infectious diseases and may have important contribution for meningococcal disease pathology. Finally, we described a novel genetic association approach that could be applied to other phenotypes.

Angioplasty in asymptomatic carotid artery stenosis vs. endarterectomy compared to best medical treatment: One-year interim results of SPACE-2

BACKGROUND

Treatment of individuals with asymptomatic carotid artery stenosis is still handled controversially. Recommendations for treatment of asymptomatic carotid stenosis with carotid endarterectomy (CEA) are based on trials having recruited patients more than 15 years ago. Registry data indicate that advances in best medical treatment (BMT) may lead to a markedly decreasing risk of stroke in asymptomatic carotid stenosis. The aim of the SPACE-2 trial (ISRCTN78592017) was to compare the stroke preventive effects of BMT alone with that of BMT in combination with CEA or carotid artery stenting (CAS), respectively, in patients with asymptomatic carotid artery stenosis of ≥70% European Carotid Surgery Trial (ECST) criteria.

METHODS

SPACE-2 is a randomized, controlled, multicenter, open study. A major secondary endpoint was the cumulative rate of any stroke (ischemic or hemorrhagic) or death from any cause within 30 days plus an ipsilateral ischemic stroke within one year of follow-up. Safety was assessed as the rate of any stroke and death from any cause within 30 days after CEA or CAS. Protocol changes had to be implemented. The results on the one-year period after treatment are reported.

FINDINGS

It was planned to enroll 3550 patients. Due to low recruitment, the enrollment of patients was stopped prematurely after randomization of 513 patients in 36 centers to CEA (n = 203), CAS (n = 197), or BMT (n = 113). The one-year rate of the major secondary endpoint did not significantly differ between groups (CEA 2.5%, CAS 3.0%, BMT 0.9%; p = 0.530) as well as rates of any stroke (CEA 3.9%, CAS 4.1%, BMT 0.9%; p = 0.256) and all-cause mortality (CEA 2.5%, CAS 1.0%, BMT 3.5%; p = 0.304). About half of all strokes occurred in the peri-interventional period. Higher albeit statistically non-significant rates of restenosis occurred in the stenting group (CEA 2.0% vs. CAS 5.6%; p = 0.068) without evidence of increased stroke rates.

INTERPRETATION

The low sample size of this prematurely stopped trial of 513 patients implies that its power is not sufficient to show that CEA or CAS is superior to a modern medical therapy (BMT) in the primary prevention of ischemic stroke in patients with an asymptomatic carotid stenosis up to one year after treatment. Also, no evidence for differences in safety between CAS and CEA during the first year after treatment could be derived. Follow-up will be performed up to five years. Data may be used for pooled analysis with ongoing trials.

Impact of suggestion on the human experimental model of cold hyperalgesia after topical application of high-concentration menthol [40%]

BACKGROUND

Human experimental pain models in healthy subjects offer unique possibilities to study mechanisms of pain within a defined setting of expected pain symptoms, signs and mechanisms. Previous trials in healthy subjects demonstrated that topical application of 40% menthol is suitable to induce cold hyperalgesia. The objective of this study was to evaluate the impact of suggestion on this experimental human pain model.

METHODS

The study was performed within a single-centre, randomized, placebo-controlled, double-blind, two-period crossover trial in a cohort of 16 healthy subjects. Subjects were tested twice after topical menthol application (40% dissolved in ethanol) and twice after ethanol (as placebo) application. In the style of a balanced placebo trial design, the subjects received during half of the testing the correct information about the applied substance (topical menthol or ethanol) and during half of the testing the incorrect information, leading to four tested conditions (treatment conditions: menthol-told-menthol and menthol-told-ethanol; placebo conditions: ethanol-told-menthol and ethanol-told-ethanol).

RESULTS

Cold but not mechanical hyperalgesia was reliably induced by the model. The cold pain threshold decreased in both treatment conditions regardless whether true or false information was given. Minor suggestion effects were found in subjects with prior ethanol application.

CONCLUSIONS

The menthol model is a reliable, nonsuggestible model to induce cold hyperalgesia. Mechanical hyperalgesia is not as reliable to induce.

SIGNIFICANCE

Cold hyperalgesia may be investigated under unbiased and suggestion-free conditions using the menthol model of pain.

Evaluating the Visualization of What a Deep Neural Network Has Learned

Deep neural networks (DNNs) have demonstrated impressive performance in complex machine learning tasks such as image classification or speech recognition. However, due to their multilayer nonlinear structure, they are not transparent, i.e., it is hard to grasp what makes them arrive at a particular classification or recognition decision, given a new unseen data sample. Recently, several approaches have been proposed enabling one to understand and interpret the reasoning embodied in a DNN for a single test image. These methods quantify the "importance" of individual pixels with respect to the classification decision and allow a visualization in terms of a heatmap in pixel/input space. While the usefulness of heatmaps can be judged subjectively by a human, an objective quality measure is missing. In this paper, we present a general methodology based on region perturbation for evaluating ordered collections of pixels such as heatmaps. We compare heatmaps computed by three different methods on the SUN397, ILSVRC2012, and MIT Places data sets. Our main result is that the recently proposed layer-wise relevance propagation algorithm qualitatively and quantitatively provides a better explanation of what made a DNN arrive at a particular classification decision than the sensitivity-based approach or the deconvolution method. We provide theoretical arguments to explain this result and discuss its practical implications. Finally, we investigate the use of heatmaps for unsupervised assessment of the neural network performance.Deep neural networks (DNNs) have demonstrated impressive performance in complex machine learning tasks such as image classification or speech recognition. However, due to their multilayer nonlinear structure, they are not transparent, i.e., it is hard to grasp what makes them arrive at a particular classification or recognition decision, given a new unseen data sample. Recently, several approaches have been proposed enabling one to understand and interpret the reasoning embodied in a DNN for a single test image. These methods quantify the "importance" of individual pixels with respect to the classification decision and allow a visualization in terms of a heatmap in pixel/input space. While the usefulness of heatmaps can be judged subjectively by a human, an objective quality measure is missing. In this paper, we present a general methodology based on region perturbation for evaluating ordered collections of pixels such as heatmaps. We compare heatmaps computed by three different methods on the SUN397, ILSVRC2012, and MIT Places data sets. Our main result is that the recently proposed layer-wise relevance propagation algorithm qualitatively and quantitatively provides a better explanation of what made a DNN arrive at a particular classification decision than the sensitivity-based approach or the deconvolution method. We provide theoretical arguments to explain this result and discuss its practical implications. Finally, we investigate the use of heatmaps for unsupervised assessment of the neural network performance.

Genetic association study of exfoliation syndrome identifies a protective rare variant at LOXL1 and five new susceptibility loci

Exfoliation syndrome (XFS) is the most common known risk factor for secondary glaucoma and a major cause of blindness worldwide. Variants in two genes, LOXL1 and CACNA1A, have previously been associated with XFS. To further elucidate the genetic basis of XFS, we collected a global sample of XFS cases to refine the association at LOXL1, which previously showed inconsistent results across populations, and to identify new variants associated with XFS. We identified a rare protective allele at LOXL1 (p.Phe407, odds ratio (OR) = 25, P = 2.9 × 10^-14) through deep resequencing of XFS cases and controls from nine countries. A genome-wide association study (GWAS) of XFS cases and controls from 24 countries followed by replication in 18 countries identified seven genome-wide significant loci (P < 5 × 10^-8). We identified association signals at 13q12 (POMP), 11q23.3 (TMEM136), 6p21 (AGPAT1), 3p24 (RBMS3) and 5q23 (near SEMA6A). These findings provide biological insights into the pathology of XFS and highlight a potential role for naturally occurring rare LOXL1 variants in disease biology.

Prediction of response to tapentadol in chronic low back pain

BACKGROUND

Many chronic low back pain (cLBP) patients do not satisfactorily respond to treatment. The knowledge of responders and non-responders before initiating treatment would improve decision making and reduce health care costs. The aims of this exploratory prediction study in cLBP patients treated with tapentadol were to identify predictors of treatment outcome based on baseline characteristics, to evaluate quality-of-life and functionality as alternative outcome parameters and to develop nomograms to calculate the individual probability of response.

METHODS

In a retrospective analysis of an open-label phase 3b trial, 46 baseline characteristics were included into statistical prediction modelling. One hundred and twenty-one patients were followed up during the titration and treatment period and 67 patients were analysed who discontinued the trial.

RESULTS

Demographic data were not relevant for response prediction. Nine baseline co-variables were robust: painDETECT score, intensity of burning and painful attacks, SF36 Health Survey score (MCS, PCS), EuroQol-5, Hospital Anxiety/Depression Scale. Gender had a minor influence. Alternative outcomes (quality-of-life, functionality) were more important for response prediction than conventional pain intensity measures. Neuropathic symptoms (high painDETECT score) had a positive predictive validity. Painful attacks and classical yellow flags (depression, anxiety) negatively influenced the treatment response. High depression scores, female gender and low burning predicted discontinuation during titration.

CONCLUSION

In this exploratory study, predictive baseline characteristics have been identified that can be used to calculate the individual probability of tapentadol response in cLBP. The small sample size in relation to the number of initial variables is a limitation of this approach.

SIGNIFICANCE

Predictors for treatment response of tapentadol were identified in patients with chronic low back pain based on clinical pre-treatment characteristics that can guide personalized treatment. Quality-of-life and functionality were the most relevant outcomes for response prediction.

Neuropathic low back pain in clinical practice

Background and objective

Low back pain (LBP) is one of the most common chronic pain conditions. This paper reviews the available literature on the role of neuropathic mechanisms in chronic LBP and discusses implications for its clinical management, with a particular focus on pharmacological treatments.

Databases and data treatment

Literature searches were performed in PubMed, key pain congresses and ProQuest Dialog to identify published evidence on neuropathic back pain and its management. All titles were assessed for relevant literature.

Results

Chronic LBP comprises both nociceptive and neuropathic components, however, the neuropathic component appears under‐recognized and undertreated. Neuropathic pain (NP) is challenging to manage. Many patients with chronic LBP have pain that is refractory to existing treatments. Typically, less than half of patients experience clinically meaningful analgesia with oral pharmacotherapies; these are also associated with risks of adverse effects. Paracetamol and NSAIDs, although widely used for LBP, are unlikely to ameliorate the neuropathic component and data on the use of NP medications such as antidepressants and gabapentin/pregabalin are limited. While there is an unmet need for improved treatment options, recent data have shown tapentadol to have efficacy in the neuropathic component of LBP, and studies suggest that the capsaicin 8% patch and lidocaine 5% medicated plaster, topical analgesics available for the treatment of peripheral NP, may be a valuable additional approach for the management of neuropathic LBP.

Conclusions

Chronic LBP often has an under‐recognized neuropathic component, which can be challenging to manage, and requires improved understanding and better diagnosis and treatment.

What does this review add?

Increased recognition and improved understanding of the neuropathic component of low back pain raises the potential for the development of mechanism‐based therapies.

Open and retrospective studies suggest that agents like tapentadol and topical analgesics — such as the capsaicin 8% patch and the lidocaine 5% medicated plaster — may be effective options for the treatment of neuropathic low back pain in defined patient groups.

[Findings in uteri and ovaries from Eringer cows slaughtered due to fertility problems]

Eringer cows are often slaughtered due to fertility problems which result from inflammatory and degenerative changes of the uterus or hormonal imbalances. Twenty-one genital tracts from Eringer cows suffering from fertility problems were collected in the abattoir. The purpose of the study was the macroscopic evaluation of the ovaries and the uterus followed by a histological and microbiological analysis of the uterus. Data from inseminations and calvings were provided by the Eringer breeding association and through the internet portal www.agate.ch. Median age of the cows was 6.9 years, number of calves per cow was 2.5 and median period between last calving and slaughter was 1.5 years. In 13 from 21 of the urogenital tracts examined, macroscopic abnormalities of the ovaries and/or histologic or microbiologic findings in the uterus could explain fertility-associated slaughter.

On Pixel-Wise Explanations for Non-Linear Classifier Decisions by Layer-Wise Relevance Propagation

Understanding and interpreting classification decisions of automated image classification systems is of high value in many applications, as it allows to verify the reasoning of the system and provides additional information to the human expert. Although machine learning methods are solving very successfully a plethora of tasks, they have in most cases the disadvantage of acting as a black box, not providing any information about what made them arrive at a particular decision. This work proposes a general solution to the problem of understanding classification decisions by pixel-wise decomposition of nonlinear classifiers. We introduce a methodology that allows to visualize the contributions of single pixels to predictions for kernel-based classifiers over Bag of Words features and for multilayered neural networks. These pixel contributions can be visualized as heatmaps and are provided to a human expert who can intuitively not only verify the validity of the classification decision, but also focus further analysis on regions of potential interest. We evaluate our method for classifiers trained on PASCAL VOC 2009 images, synthetic image data containing geometric shapes, the MNIST handwritten digits data set and for the pre-trained ImageNet model available as part of the Caffe open source package.

Thrombectomy vs. Systemic Thrombolysis in Acute Embolic Stroke with High Clot Burden: A Retrospective Analysis

PURPOSE

The efficacy of i. v. thrombolysis in acute stroke with high clot burden is limited. Successful recanalization is very unlikely if the thrombus length exceeds 7 mm. Thus this retrospective controlled study evaluated the efficacy and safety of neurothrombectomy in the treatment of acute embolic stroke in patients selected by a thrombus length of ≥ 8 mm using the stent retriever Trevo(®) device.

MATERIALS AND METHODS

40 patients with acute occlusion of the anterior intracranial arteries with a thrombus length of ≥ 8 mm were treated with neurothrombectomy. We compared the outcome with a historical cohort of 42 patients with a thrombus length of ≥ 8 mm that received i. v. thrombolysis only. Clinical outcome was assessed by modified Rankin scale in both groups at discharge and on day 90.

RESULTS

Patients did not differ in age, mRS on admission, thrombus length or time from symptom onset to i. v. thrombolysis, but the thrombectomy group had higher NIHSS on admission. Successful recanalization was achieved in 33/40 patients (83 %) with neurothrombectomy. 15 patients received i. v. thrombolysis prior to neurothrombectomy. Median mRS at discharge was 3.5 (1.25 - 5) vs. 5 (4 - 6; p < 0.01) and on day 90 3 (1 - 4) vs. 5 (4 - 6; p < 0.01). Symptomatic hemorrhage occurred in 3 vs. 7 patients. 3 vs. 17 patients died within 90 days (thrombectomy vs. control each). There were only a few intervention-related complications.

CONCLUSION

Thrombectomy in acute stroke with high clot burden using the Trevo(®) device has a low risk and improved clinical outcome compared to i. v. thrombolysis alone. Treatment selection by a clot length of ≥ 8 mm might be a powerful approach to improve the outcome of mechanical thrombectomy.

KEY POINTS

• Clot length of ≥ 8 mm might be a valuable criterion for indicating neurothrombectomy. • Thrombolysis only in high clot burden is associated with poor clinical outcome. • Thrombectomy using the Trevo(®) stent retriever is safe and effective.

[Reembursing health-care service provider networks]

Health-care service provider networks are regarded as an important instrument to overcome the widely criticised fragmentation and sectoral partition of the German health-care system. The first part of this paper incorporates health-care service provider networks in the field of health-care research. The system theoretical model and basic functions of health-care research are used for this purpose. Furthermore already established areas of health-care research with strong relations to health-care service provider networks are listed. The second part of this paper introduces some innovative options for reimbursing health-care service provider networks which can be regarded as some results of network-oriented health-care research. The origins are virtual budgets currently used in part to reimburse integrated care according to §§ 140a ff. SGB V. Describing and evaluating this model leads to real budgets (capitation) - a reimbursement scheme repeatedly demanded by SVR-Gesundheit (German governmental health-care advisory board), for example, however barely implemented. As a final step a direct reimbursement of networks by the German sickness fund is discussed. Advantages and challenges are shown. The development of the different reimbursement schemes is partially based on models from the USA.

Topical high-concentration menthol: reproducibility of a human surrogate pain model

BACKGROUND

Human experimental pain models play an important role in studying neuropathic pain mechanisms. The objective of the present study was to test the reproducibility of the topical menthol model over a 1-week period.

METHOD

We performed an open, two-period study in 10 healthy volunteers with 40 menthol applications. The side of menthol application was randomly assigned. Two trial periods were separated by 1 week. Before and after applying menthol, selected quantitative sensory testing (QST) was performed. The area of mechanical pin-prick hyperalgesia was quantified. Spontaneous pain was recorded.

RESULTS

Application of menthol induced a statistically significant decrease in the cold pain threshold (CPT) (p < 0.001) and mechanical pain threshold and an increase in the mechanical pain sensitivity (MPS) (p < 0.001), indicating cold and mechanical (pin-prick) hyperalgesia. Test-retest reliability was best for CPT (r = 0.959) and MPS (r = 0.930). Intraclass correlation values showed excellent reliability for cold pain and MPS (ICC = 0.96, 0.89). The QST values post-menthol showed high inter-period correlation factors and no significant inter-period differences (paired t-test, t = 1.767-1.361; p = 0.111-0.988). The area size of mechanical hyperalgesia was not reliably reproducible.

CONCLUSION

For an observation period of 1 week, the signs of cold and mechanical hyperalgesia were reproducible with a highly significant correlation of about r = 0.8 and good agreement except for the area size of mechanical pin-prick hyperalgesia. These results demonstrate that the topical menthol pain model is suitable for pharmacological interventions repeated within an observation period of 1 week.

Multiple kernel learning for brain-computer interfacing

Combining information from different sources is a common way to improve classification accuracy in Brain-Computer Interfacing (BCI). For instance, in small sample settings it is useful to integrate data from other subjects or sessions in order to improve the estimation quality of the spatial filters or the classifier. Since data from different subjects may show large variability, it is crucial to weight the contributions according to importance. Many multi-subject learning algorithms determine the optimal weighting in a separate step by using heuristics, however, without ensuring that the selected weights are optimal with respect to classification. In this work we apply Multiple Kernel Learning (MKL) to this problem. MKL has been widely used for feature fusion in computer vision and allows to simultaneously learn the classifier and the optimal weighting. We compare the MKL method to two baseline approaches and investigate the reasons for performance improvement.

A new scoring system derived from base excess and platelet count at presentation predicts mortality in paediatric meningococcal sepsis

INTRODUCTION

The aim of this study was to derive a novel prognostic score for mortality in paediatric meningococcal sepsis (MS) based on readily available laboratory markers.

METHODS

A multicentre retrospective cohort study for the consortium set and a single centre retrospective study for replication set. The consortium set were 1,073 children (age 1 week to 17.9 years) referred over a 15-year period (1996 to 2011), who had an admission diagnosis of MS, referred to paediatric intensive care units (PICUs) in six different European centres. The consortium set was split into a development set and validation set to derive the score. The replication set were 134 children with MS (age 2 weeks to 16 years) referred over a 4-year period (2007 to 2011) to PICUs via the Children's Acute Transport Service (CATS), London.

RESULTS

A total of 85/1,073 (7.9%) children in the consortium set died. A total of 16/134 (11.9%) children in the replication set died. Children dying in the consortium set had significantly lower base excess, C-reactive protein (CRP), platelet and white cell count, more deranged coagulation and higher lactate than survivors. Paediatric risk of mortality (PRISM) score, Glasgow meningococcal septicaemia prognosis score (GMSPS) and Rotterdam score were also higher. Using the consortium set, a new scoring system using base excess and platelet count at presentation, termed the BEP score, was mathematically developed and validated. BEP predicted mortality with high sensitivity and specificity scores (area under the curve (AUC) in the validation set=0.86 and in the replication set=0.96). In the validation set, BEP score performance (AUC=0.86, confidence interval (CI): 0.80 to 0.91) was better than GMSPS (AUC=0.77, CI: 0.68, 0.85), similar to Rotterdam (AUC=0.87, CI: 0.81 to 0.93) and not as good as PRISM (AUC=0.93, CI: 0.85 to 0.97).

CONCLUSIONS

The BEP score, relying on only two variables that are quickly and objectively measurable and readily available at presentation, is highly sensitive and specific in predicting death from MS in childhood.

Insights from classifying visual concepts with multiple kernel learning

Combining information from various image features has become a standard technique in concept recognition tasks. However, the optimal way of fusing the resulting kernel functions is usually unknown in practical applications. Multiple kernel learning (MKL) techniques allow to determine an optimal linear combination of such similarity matrices. Classical approaches to MKL promote sparse mixtures. Unfortunately, 1-norm regularized MKL variants are often observed to be outperformed by an unweighted sum kernel. The main contributions of this paper are the following: we apply a recently developed non-sparse MKL variant to state-of-the-art concept recognition tasks from the application domain of computer vision. We provide insights on benefits and limits of non-sparse MKL and compare it against its direct competitors, the sum-kernel SVM and sparse MKL. We report empirical results for the PASCAL VOC 2009 Classification and ImageCLEF2010 Photo Annotation challenge data sets. Data sets (kernel matrices) as well as further information are available at http://doc.ml.tu-berlin.de/image_mkl/(Accessed 2012 Jun 25).

Quantitative sensory testing in the German Research Network on Neuropathic Pain (DFNS): somatosensory abnormalities in 1236 patients with different neuropathic pain syndromes

Neuropathic pain is accompanied by both positive and negative sensory signs. To explore the spectrum of sensory abnormalities, 1236 patients with a clinical diagnosis of neuropathic pain were assessed by quantitative sensory testing (QST) following the protocol of DFNS (German Research Network on Neuropathic Pain), using both thermal and mechanical nociceptive as well as non-nociceptive stimuli. Data distributions showed a systematic shift to hyperalgesia for nociceptive, and to hypoesthesia for non-nociceptive parameters. Across all parameters, 92% of the patients presented at least one abnormality. Thermosensory or mechanical hypoesthesia (up to 41%) was more frequent than hypoalgesia (up to 18% for mechanical stimuli). Mechanical hyperalgesias occurred more often (blunt pressure: 36%, pinprick: 29%) than thermal hyperalgesias (cold: 19%, heat: 24%), dynamic mechanical allodynia (20%), paradoxical heat sensations (18%) or enhanced wind-up (13%). Hyperesthesia was less than 5%. Every single sensory abnormality occurred in each neurological syndrome, but with different frequencies: thermal and mechanical hyperalgesias were most frequent in complex regional pain syndrome and peripheral nerve injury, allodynia in postherpetic neuralgia. In postherpetic neuralgia and in central pain, subgroups showed either mechanical hyperalgesia or mechanical hypoalgesia. The most frequent combinations of gain and loss were mixed thermal/mechanical loss without hyperalgesia (central pain and polyneuropathy), mixed loss with mechanical hyperalgesia in peripheral neuropathies, mechanical hyperalgesia without any loss in trigeminal neuralgia. Thus, somatosensory profiles with different combinations of loss and gain are shared across the major neuropathic pain syndromes. The characterization of underlying mechanisms will be needed to make a mechanism-based classification feasible.

Maternal interleukin-6 (-174) C/C polymorphism is associated with chorioamnionitis and cystic periventricular leucomalacia of the preterm infant

OBJECTIVE

To evaluate the association between maternal interleukin (IL)-6 G(-174)C polymorphism and cystic periventricular leukomalacia (cPVL) of the preterm newborn.

STUDY DESIGN

After searching a local database, we recruited 132 preterm infants with diagnosis of cPVL, 44 Caucasian mothers were also recruited to participate in this candidate gene-association study at a single teritary care center. Data related to maternal IL-6 G(-174)C polymorphisms were compared with 41 controls, and furthermore compared with data from umbilical cord blood samples from a consecutive birth cohort of 395 healthy newborns, and published data from Caucasian populations including 1104 adults, respectively. In addition, subgroup analysis was performed in cases with either history of preterm premature rupture of the membranes (PPROM) or clinical chorioamnionitis (CCA). IL-6 genotyping was performed using an allele-specific polymerase chain reaction technique.

RESULT

Frequencies of the IL-6 G(-174)C polymorphisms did not differ between cases (GG, 29.5%; GC, 54.5% and CC, 15.9%) and controls (GG, 34.2; GC, 51.2 and CC, 14.6%). Subgroup analysis of 31 cases with history of PPROM (GG, 25.8; GC, 54.8 and CC 19.4%) and controls did not reveal significant differences, but a significantly higher frequency of the CC genotype was found in 23 cases with a history of CCA (34.8%) compared with controls by either univariate (P=0.032; odds ratio 3.11, 95% confidence interval (CI) 1.11 to 8.68) or multivariate analysis (P=0.049, odds ratio 2.54, 95% CI 1.01 to 6.45). These data were confirmed by a comparing the CC genotype frequency to 395 term controls (CC 14.7%, P=0.005) and to the mean CC genotype frequency of 1104 Caucasian adults (CC 15.6%, P<0.0001).

CONCLUSION

Frequencies of the IL-6 G(-174)C polymorphisms did not differ between groups. Subgroup analysis revealed an association of the CC genotype with CCA and cPVL in the preterm newborn.

Photosystem I reaction center from the thermophilic cyanobacterium Mastigocladus laminosus

Photosystem I reaction center was isolated from the cyanobacterium Mastigocladus laminosus. It contained four different subunits with molecular masses (as determined by sodium dodecyl sulfate gels) of about 70,000 (subunit I), 16,000 (subunit II), 11,000 (subunit III), and 10,000 (subunit IV) daltons. The purified reaction center contained about 100 chlorophyll a molecules per P(700); however, they could be readily depleted down to about 50 chlorophyll a per P(700) without loss in the photochemical activities. The reaction center was active in cytochrome c photooxidation, but the photooxidation of an acidic cytochrome, like the Euglena cytochrome 552, required the presence of cations. The purified reaction center was found to be similar in several respects to the photosystem I reaction centers from higher plants and, especially, to the one isolated from green algae. Subunit I appeared on sodium dodecyl sulfate gels in the same position and possessed the same shape of an apparent double band as the corresponding subunits I of green plants and of algae. Subunits I and II of photosystem I reaction centers from Mastigocladus, higher plants, and green algae showed immunological crossreactivity. This observation might serve as biochemical evidence for the common evolution of the photosystem I reaction centers. In higher plants and green algae subunit II is a product of cytoplasmic ribosomes and therefore, a high degree of homology should have been preserved upon transfer of its gene from the prokaryote to the nucleus of the eukaryotes.