VACS: VAccination disComfort Scale.

The vaccination of children is a crucial tool to protect both individuals and the world in general from various diseases and pathogens. Unfortunately, the vaccination procedure is not a pleasant one for all children, with many experiencing various levels of discomfort, sometimes reaching intolerable levels. In the first part of this work, we develop VACS, a tool that measures the discomfort children experience during vaccination. VACS takes into consideration the complete timeline of the vaccination experience from the perspective of the child, starting from the moment the child enters the doctor's office through to their departure, and also the complete range of manifestations of discomfort, ranging from moaning and crying to facial expressions and posture. Their discomfort is quantified as a number from 0 to 25, with zero corresponding to a smooth vaccination and 25 to maximal/unbearable discomfort. In the second part of the work, we apply VACS to 40 vaccinations of children aged 2 to 12. Our findings show that approximately 40% of the children do not face discomfort during vaccination, but for the rest discomfort of varying degrees is observed. We also find that doctors are content with their patients facing considerably higher discomfort levels than what the children themselves are willing to withstand: doctors are content with VACS values up to 19 whilst children start to suffer when the VACS value exceeds 11. Surprisingly, characteristics such as (a) gender, (b) whether the state's recommended vaccination program has been implemented in full, and even (c) prior negative vaccination experiences are found to be poor predictors of vaccination discomfort. Age on the other hand may be a factor, with younger children experiencing discomfort more often and more intensely; more research is required in order to validate this with higher confidence. The formulation of VACS opens the door for more systematic work towards the mitigation of vaccination discomfort for children.

Utilization of evidence theory in the detection of salient regions in successive CT images.

This study presents an integrated approach to locating and presenting the medical practitioner with salient regions in a computed tomography (CT) scan when focusing on the area of the liver. A number of image processing tasks are performed in successive scans to extract areas with a different features than that of the greater part of the organ. In general, these areas do not always correspond to pathological patterns, but may be the result of noise in the scanned image or related to veins passing through the tissue. The result of the algorithm is the original image with a mask indicating these regions, so the attention of the medical practitioner is drawn to them for further examination. The algorithm also calculates a measure of confidence of the system, with respect to the extraction of the salient region, based on the fact that a region with a similar pattern is also located in successive scans. This essentially represents the hypothesis that the volume of both pathological patterns and blood vessels, but not noise patterns, is large enough to be captured in successive scans.

Intelligent initialization of resource allocating RBF networks.

In any neural network system, proper parameter initialization reduces training time and effort, and generally leads to compact modeling of the process under examination, i.e. less complex network structures and better generalization. However, in cases of multi-dimensional data, parameter initialization is both difficult and time consuming. In the proposed scheme a novel, multi-dimensional, unsupervised clustering method is used to properly initialize neural network architectures, focusing on resource allocating networks (RAN); both the hidden and output layer parameters are determined by the output of the clustering process, without the need for any user interference. The main contribution of this work is that the proposed approach leads to network structures that are compact, efficient and achieve best classification results, without the need for manual selection of suitable initial network parameters. The efficiency of the proposed method has been tested on several classes of publicly available data, such as iris, Wisconsin and ionosphere data.