Reinforcement learning in dendritic structures

The discovery of binary dendritic events such as local NMDA spikes in dendritic subbranches led to the suggestion that dendritic trees could be computationally equivalent to a 2-layer network of point neurons, with a single output unit represented by the soma, and input units represented by the dendritic branches. Although this interpretation endows a neuron with a high computational power, it is functionally not clear why nature would have preferred the dendritic solution with a single but complex neuron, as opposed to the network solution with many but simple units. We show that the dendritic solution has a distinguished advantage over the network solution when considering different learning tasks. Its key property is that the dendritic branches receive an immediate feedback from the somatic output spike, while in the corresponding network architecture the feedback would require additional backpropagating connections to the input units. Assuming a reinforcement learning scenario we formally derive a learning rule for the synaptic contacts on the individual dendritic trees which depends on the presynaptic activity, the local NMDA spikes, the somatic action potential, and a delayed reinforcement signal. We test the model for two scenarios: the learning of binary classifications and of precise spike timings. We show that the immediate feedback represented by the backpropagating action potential supplies the individual dendritic branches with enough information to efficiently adapt their synapses and to speed up the learning process.

Reinforcement learning in dendritic structures

The discovery of binary dendritic events such as local NMDA spikes in dendritic subbranches led to the suggestion that dendritic trees could be computationally equivalent to a 2-layer network of point neurons, with a single output unit represented by the soma, and input units represented by the dendritic branches. Although this interpretation endows a neuron with a high computational power, it is functionally not clear why nature would have preferred the dendritic solution with a single but complex neuron, as opposed to the network solution with many but simple units. We show that the dendritic solution has a distinguished advantage over the network solution when considering different learning tasks. Its key property is that the dendritic branches receive an immediate feedback from the somatic output spike, while in the corresponding network architecture the feedback would require additional backpropagating connections to the input units. Assuming a reinforcement learning scenario we formally derive a learning rule for the synaptic contacts on the individual dendritic trees which depends on the presynaptic activity, the local NMDA spikes, the somatic action potential, and a delayed reinforcement signal. We test the model for two scenarios: the learning of binary classifications and of precise spike timings. We show that the immediate feedback represented by the backpropagating action potential supplies the individual dendritic branches with enough information to efficiently adapt their synapses and to speed up the learning process.

Measuring attitudes, behaviours, and influences in inner city victims of interpersonal violence (VIVs) - a Swiss emergency room pilot study

Background Switzerland is confronted with the problem of interpersonal violence. Violence is in the increase and the potential for aggression seems to be rising. Observations by hospitals discern an appalling increase of the severity of the injuries. The aim of this study is to collect accurate information about the social environment, the motivation and possible reasons for violence. We also intend to investigate whether sociocultural, or ethnic differences among male victims exist. Materials and methods For the first time in Switzerland, this survey employed a validated questionnaire from the division of violence prevention, Atlanta, Georgia. The first part of the questionnaire addressed social and demographic factors which could influence the risk of violence: age, gender, duration of stay in Switzerland, nationality and educational level. Beside these social structural factors, the questionnaire included questions on experience of violent offences in the past, information about the most recent violent offence and intra and interpersonal facts. The questionnaire itself consists of 27 questions, translated into German and French. In a pilot study, the questionnaire was checked with adolescents for feasibility and comprehensibility. Results 69 male VIVs were interviewed at two hospitals in the Canton of Bern. Most of the adolescents emphasised that weapons were not used during their confrontations. It is astonishing that all of the young men considered themselves to be victims. Most of the brawls were incited after an exchange of verbal abuse and provocations with unfamiliar individuals. The rivals could neither be classified with the help of ethnic categories nor identifiable groups of the youth scenes. The incidents took place in scenes, where violence was more likely to happen. Interestingly and contrary to a general perception the offenders are well integrated into sport and leisure clubs. A further surprising result of our research is that the attitude towards religion differs between young men with experience of violence and non-violent men. Discussion Youth violence is a health issue, which concerns us globally. The human and economic toll of violence on victims and offenders, their families, and on society in general is high. The economic costs associated with violence-related illness and disability is estimated to be millions of Swiss francs each year. Physicians and psychologists are compelled to identify the factors, which cause young people to be violent, to find out which interventions prove to be successful, and to design effective prevention programs. The identification of effective programs depends on the availability of reliable and valid measures to assess changes in violence-related attitudes. In our efforts to create healthier communities, we need to investigate; document and do research on the causes and circumstances of youth violence.

Delivery room management of very low birth weight infants in Germany, Austria and Switzerland--a comparison of protocols

Surveys from the USA, Australia and Spain have shown significant inter-institutional variation in delivery room (DR) management of very low birth weight infants (VLBWI, <1500g) at birth, despite regularly updated international guidelines.

Dysnatraemias in the emergency room: Undetected, untreated, unknown?

Hypo- and hypernatraemia are the most common electrolyte disorders in hospitalized patients and have been associated with increased mortality. However, data on the prevalence of dysnatraemias in the emergency room and the characteristics of patients presenting with them are rare.

Ordering CT pulmonary angiography to exclude pulmonary embolism: defense versus evidence in the emergency room

To identify reasons for ordering computed tomography pulmonary angiography (CTPA), to identify the frequency of reasons for CTPA reflecting defensive behavior and evidence-based behavior, and to identify the impact of defensive medicine and of training about diagnosing pulmonary embolism (PE) on positive results of CTPA.

Gradient estimation in dendritic reinforcement learning

We study synaptic plasticity in a complex neuronal cell model where NMDA-spikes can arise in certain dendritic zones. In the context of reinforcement learning, two kinds of plasticity rules are derived, zone reinforcement (ZR) and cell reinforcement (CR), which both optimize the expected reward by stochastic gradient ascent. For ZR, the synaptic plasticity response to the external reward signal is modulated exclusively by quantities which are local to the NMDA-spike initiation zone in which the synapse is situated. CR, in addition, uses nonlocal feedback from the soma of the cell, provided by mechanisms such as the backpropagating action potential. Simulation results show that, compared to ZR, the use of nonlocal feedback in CR can drastically enhance learning performance. We suggest that the availability of nonlocal feedback for learning is a key advantage of complex neurons over networks of simple point neurons, which have previously been found to be largely equivalent with regard to computational capability.

Prevalence and associated factors of viral hepatitis and transferrin elevations in 5036 patients admitted to the emergency room of a Swiss university hospital: cross-sectional study

BACKGROUND: The epidemiology of liver disease in patients admitted to emergency rooms is largely unknown. The current study aimed to measure the prevalence of viral hepatitis B and C infection and pathological laboratory values of liver disease in such a population, and to study factors associated with these measurements. METHODS: Cross-sectional study in patients admitted to the emergency room of a university hospital. No formal exclusion criteria. Determination of anti-HBs, anti-HCV, transferrin saturation, alanine aminotransferase, and obtaining answers from a study-specific questionnaire. RESULTS: The study included 5'036 patients, representing a 14.9% sample of the target population during the study period. Prevalence of anti-HBc and anti-HCV was 6.7% (95%CI 6.0% to 7.4%) and 2.7% (2.3% to 3.2%), respectively. Factors independently associated with positive anti-HBc were intravenous drug abuse (OR 18.3; 11.3 to 29.7), foreign country of birth (3.4; 2.6 to 4.4), non-white ethnicity (2.7; 1.9 to 3.8) and age > or =60 (2.0; 1.5 to 2.8). Positive anti-HCV was associated with intravenous drug abuse (78.9; 43.4 to 143.6), blood transfusion (1.7; 1.1 to 2.8) and abdominal pain (2.7; 1.5 to 4.8). 75% of all participants were not vaccinated against hepatitis B or did not know their vaccination status. Among anti-HCV positive patients only 49% knew about their infection and 51% reported regular alcohol consumption. Transferrin saturation was elevated in 3.3% and was associated with fatigue (prevalence ratio 1.9; 1.2 to 2.8). CONCLUSION: Emergency rooms should be considered as targets for public health programs that encourage vaccination, patient education and screening of high-risk patients for liver disease with subsequent referral for treatment if indicated.

Minimum ten year results of total hip arthroplasty with the acetabular reinforcement ring in avascular osteonecrosis

Total hip arthroplasty (THA) still carries a higher failure rate in patients with avascular necrosis of the femoral head (AVN) than in a similar patient population with THA for other reasons. This is particularly true for the acetabular component. One of the major factors accounting for this is the compromised acetabular bone quality with structural defects subsequent to collapsing of the femoral head in high-grade AVN. In this study we implanted an acetabular reinforcement ring with hook (ARRH), which had been used successfully for other indications with acetabular bone stock deficiency, in 32 consecutive THA's in 29 patients with AVN. Five patients died during the observation period of causes unrelated to the surgery, one patient was lost to follow-up and one patient could not be followed up due to chronic illness, leaving 25 hips (23 patients) with a minimum follow-up of ten years (mean: 11.8; range: 10-15). The mean Merle d'Aubigne score increased significantly from 7.7 preoperatively to 16.6 postoperatively (p < 0.001). One revision was performed for aseptic stem loosening. Of the unrevised hips, one acetabular component was classified as definitively loose. The cumulative 12-year survivorship for THA with ARRH in AVN was 95.2% (confidence interval: 86.1-100%) for both components, 100% for the cup and 95.2% for the stem (86.1-100%).