Britta Kehden

Attributable risks for childhood overweight: evidence for limited effectiveness of prevention

OBJECTIVE: Calculation of attributable risks (ARs) of childhood overweight to estimate effectiveness of prevention strategies. METHODS: We used pooled data of 4 population-based German studies including 34 240 children and adolescents aged 3 to 18 years to calculate the impact of familial, social, “early life”, and lifestyle factors on overweight. ARs (joint for all determinants as well as partial risks) were calculated. RESULTS: The prevalence of childhood overweight was 13.4%. Successfully tackling all determinants can reduce overweight by 77.7% (ie, from 13.4% to 3.0%; = joint AR) with partial effects of treating parental overweight (42.5%); improving social status (14.3%); reducing media time to <1 hour per day (11.4%); and not smoking during pregnancy, low weight gain during pregnancy, and breastfeeding (together 9.5%), respectively. Improving all preventable risk factors (ie, early life factors and lifestyle) the effect is 9.2%. Media time has the strongest effect. CONCLUSIONS: The determinants identified explained 78% of the prevalence of overweight. Taking into account the partial ARs, the effectiveness of lifestyle interventions to prevent overweight in children is limited. Our data argue in favor of interventions aimed at families and social environments, with a major focus on promoting a lower screen time and computer use in children.

Body Fat Percentiles for German Children and Adolescents

Objective: The aim of this study was to define body fat percentiles for German children and adolescents aged 3–16 years using the largest German database. Methods: The study population included 11,632 girls and 11,604 boys. Data were pooled from: i) Kiel Obesity Prevention Study (KOPS), acquisition period: 1996–2008, n = 12,237; ii) ‘Better diet. More exercise. KINDERLEICHT-REGIONS’, acquisition period: 2007, n = 9,405; and iii) examination of Jena schoolchildren, acquisition period: 2005, n = 1,594. Body fat mass was measured by bioelectrical impedance analysis using a population-specific algorithm. Data were weighted to achieve a representative sample for Germany. Percentile curves were constructed by the LMS method and proved by Worm plots and Q-statistic. Results: In both genders, the higher body fat percentile curves sloped downwards to age 7 years, whereas the lower percentiles declined up to 8.5 years. Thereafter fat mass remained nearly constant with age in boys and increased in girls. The 10th percentile achieved a minimum of 10–11% body fat in both genders, whereas the 90th percentile curve fluctuated between 29 and 44% in boys or 30–43% in girls. The association between fat mass and blood pressure was too weak to define disease-related cut-offs. Conclusion: These body fat percentiles are suitable reference values for German children and adolescents.

State space analysis of Petri nets with relation-algebraic methods

A large variety of systems can be modelled by Petri nets. Their formal semantics are based on linear algebra which in particular allows the calculation of a Petri nets state space. Since state space explosion is still a serious problem, efficiently calculating, representing, and analysing the state space is mandatory. We propose a formal semantics of Petri nets based on executable relation-algebraic specifications. Thereupon, we suggest how to calculate the markings reachable from a given one simultaneously. We provide an efficient representation of reachability graphs and show in a correct-by-construction approach how to efficiently analyse their properties. Therewith we cover two aspects: modelling and model checking systems by means of one and the same logic-based approach. On a practical side, we explore the power and limits of relation-algebraic concepts for concurrent system analysis.

Evaluating sets of search points using relational algebra

We model a set of search points as a relation and use relational algebra to evaluate all elements of the set in one step in order to select search points with certain properties. Therefore we transform relations into vectors and prove a formula to translate properties of relations into properties of the corresponding vectors. This approach is applied to timetable problems.

A relation-algebraic view on evolutionary algorithms for some graph problems

We take a relation-algebraic view on the formulation of evolutionary algorithms in discrete search spaces. First, we show how individuals and populations can be represented as relations and formulate some standard mutation and crossover operators for this representation using relation-algebra. Evaluating a population with respect to their constraints seems to be the most costly step in one generation for many important problems. We show that the evaluation process for a given population can be sped up by using relation-algebraic expressions in the process. This is done by examining the evaluation of possible solutions for three of the best-known NP-hard combinatorial optimization problems on graphs, namely the vertex cover problem, the computation of maximum cliques, and the determination of a maximum independent set. Extending the evaluation process for a given population to the evaluation of the whole search space we get exact methods for the considered problems, which allow to evaluate the quality of solutions obtained by evolutionary algorithms.

Relation-algebraic specification and solution of special university timetabling problems

Abstract In this paper, we are concerned with a special timetabling problem. It was posed to us by the administration of our university and stems from the adoption of the British-American system of university education in Germany. This change led to the concrete task of constructing a timetable that enables the undergraduate education of secondary school teachers within three years in the “normal case” and within four years in the case of exceptional combinations of subjects. We develop two relation-algebraic models of the timetabling problem and in each case algorithms for computing solutions. The latter easily can be implemented in the Kiel RelView tool showing that RelView can be used for timetabling.

Relational implementation of simple parallel evolutionary algorithms

Randomized search heuristics, among them evolutionary algorithms, are applied to problems whose structure is not well understood, as well as to hard problems in combinatorial optimization to get near-optimal solutions. We present a new approach implementing simple parallel evolutionary algorithms by relational methods. Populations are represented as relations which are implicitly encoded by (reduced, ordered) binary decision diagrams. Thereby, the creation and evaluation is done in parallel, which increases efficiency considerably.

Multi-objective problems in terms of relational algebra

Relational algebra has been shown to be a powerful tool for solving a wide range of combinatorial optimization problems with small computational and programming effort. The problems considered in recent years are single- objective ones where one single objective function has to be optimized. With this paper we start considerations on the use of relational algebra for multi-objective problems. In contrast to single-objective optimization multiple objective functions have to be optimized at the same time usually resulting in a set of different trade-offs with respect to the different functions. On the one hand, we examine how to solve the mentioned problem exactly by using relational algebraic programs. On the other hand, we address the problem of objective reduction that has recently been shown to be NP-hard. We propose an exact algorithm for this problem based on relational algebra. Our experimental results show that this algorithm drastically outperforms the currently best one.

Wie groß muss die Wirkung von Prävention auf das Übergewicht von Kindern und Jugendlichen sein

BACKGROUND AND AIM OF THE STUDY Successful preventive measures can stop a further increase in the prevalence of overweight in children and adolescents. However, up to now, the required effect sizes of interventions for reducing childhood overweight remain unclear. The calculation of the energy gap (excess calories consumed over calories expended) offers the possibility to estimate the required effect sizes. In this work 2 approaches to calculate the energy gap will be compared. METHODS Longitudinal data of 1690 children and adolescents of the Kiel Obesity Prevention Study (KOPS) on measured height, weight, fat mass and fat-free mass (using bioelectrical impedance analysis) at age 6 and 10 years will be used to calculate energy gap with 2 different approaches: (i) using age-independent changes in fat mass and fat-free mass (old approach) and (ii) using a mathematic model of weight dynamic (new approach). RESULTS Energy gap according to the old approach was 140 kcal/day; by contrast, new modeling resulted in an energy gap between 270 and 370 kcal/day. Both, BMI and fat mass were suitable to calculate energy gap and led to nearly same results. Exceeding the 90(th) percentile of BMI or fat mass (incidence approach) as well as large changes within the normal range (i.e. between the 10(th) and the 90(th) percentile) led to large energy gaps. Thus, all children with large energy gaps have to be characterized as at risk for overweight. CONCLUSION The new approach seems to be convincing because it considered the additional energy expenditure for building fat-free mass due to increasing age and weight.Calculating energy gap offers a new approach for prevention of overweight. It shows that the required effect sizes of prevention measures have to be in a region of 140 to 400 kcal/day. This differs clearly from energy reduction of diets in the therapy of obesity.

Relational Construction of Specific Timetables

We use relation algebra to model an abstract timetabling problem and to compute solutions. The problem was posed to us by the administration of our university and stems from the current change from the classical German university education system to the undergraduate-graduate system. In particular with regard to the undergraduate education of secondary school teachers this change causes some difficulties. A very serious one is to enable a three years duration of study without to abolish Germany’s tradition of (at least) two different fields of study. Exactly this demand is the background of our specific university timetabling problem. We will show how to transform an informal problem description into a formal relation-algebraic model. Using it as starting point, we then develop an algorithm for obtaining solutions. In essence the algorithm is given by a relation-algebraic expression that immediately can be translated into the programming language of the RelView tool. Because of the moderate size of the timetabling problem and the very efficient BDD-implementation of relations in RelView, this even allows to compute all existing solutions of the problem or to message that no solution exists. Due to space restrictions we omit many details. This material will be included in [3].