Statistical phenomena in hospital epidemiology: Challenges for statisticians and clinicians
Freitag, 22.4.16, 12:00-13:00, Raum 404, Eckerstr. 1
Communicating statistical methods and interpreting results to clinicians belong to\nthe main tasks of medical statisticians. In this talk, I discuss several results of recent\npublications in high-impactjournals about the determinants and consequences of\nhospital-acquired infections. In such time-dependent analyses, one is confronted\nwith competing and intermediate events as well as time-dependent covariates. I\nwill give special emphasis on two different metrics which are often used without\nany distinction and which can easily lead to confusion among clinicians. Further,\nI will give an outlook about unsolved challenges for medical statisticians.
Competing selective sweeps
Freitag, 6.5.16, 12:00-13:00, Raum 404, Eckerstr. 1
In population genetics, mathematical models are used to study the distributions and changes of allele frequencies. Main evolutionary factors influencing these frequencies are (among others) mutation, selection and recombination. Maynard Smith and Haigh (1974) analysed in a pioneering theoretical framework the process when a new, strongly selected advantageous mutation becomes fixed in a population. They identified that such an evolution, called selective sweep, leads to the reduction of diversity around the selective locus. In the following years other scientists faced the question to what extent this characteristic still holds, when certain assumptions are modified. \n\nIn this talk a situation is presented where two selective sweeps within a narrow genomic region overlap in a sexually evolving population. For such a competing sweeps situation the probability of a fixation of both beneficial alleles, in cases where these alleles are not initially linked, is examined. To handle this question a graphical tool, the ancestral selection recombination graph, is utilized, which is based on a genealogical view on the population. This approach provides a limit result (for large selection coefficients) for the probability that both beneficial mutations will eventually fix. The analytical examination is complemented by simulation results.
Capital allocation for dynamic risk measures
Freitag, 3.6.16, 12:00-13:00, Raum 125, Eckerstr. 1
apital allocations have been studied in conjunction with static risk measures in various\npapers. The dynamic case has been studied only in a discrete-time setting. We address the\nproblem of allocating risk capital to subportfolios in a continuous-time dynamic context.\nFor this purpose we introduce a classical differentiability result for backward stochastic\nVolterra integral equations and apply this result to derive continuous-time dynamic capital\nallocations. Moreover, we study a dynamic capital allocation principle that is based on\nbackward stochastic differential equations and derive the dynamic gradient allocation for\nthe dynamic entropic risk measure. As a consequence we finally provide a representation result for\ndynamic risk measures that is based on the full allocation property of the Aumann-Shapley\nallocation, which is also new in the static case.\n
New Concepts for Reliable Assessment of Statistical Methods
Freitag, 10.6.16, 12:00-13:00, Raum 404, Eckerstr. 1
In Bioinformatics and Systems Biology, a huge variety of computational tools and statistical approaches have been developed. However, many computational methods are not well-tested in application settings and their applicability is often seriously delimited. Therefore, selecting an optimal analysis strategy of often difficult in applications and missing guidelines hamper the transfer of theoretical approaches to experimental research.\nIn this talk, new concepts for assessing statistical algorithms will be introduced and illustrated. The suggested methodology enables less biased, more reliable and valid comparisons of competing approaches than currently performed in the literature. The presented concepts can be applied to establish optimized analysis pipelines and for developing general decision guidelines for the selection of appropriate analysis methods. Thereby, the presented methodology constitutes a promising perspective for transferring computational approaches to basic research in academia and to industrial applications like drug development.
A short trip through the tree of life: from Ebola over Diphtheria and Tuberculosis to Penguins
Freitag, 24.6.16, 12:00-13:00, Raum 404, Eckerstr. 1
Genetic sequencing data contain a fingerprint of past evolutionary and population dynamic processes. Phylogenetic methods infer evolutionary relationships — the phylogenetic tree — between individuals based on their genetic sequences. Phylodynamics aims to understand the population dynamic processes — such as epidemiological or macroevolutionary processes — giving rise to the phylogenetic tree. I will present the mathematical and computational aspects of our recently developed phylodynamic tools. Then I will discuss epidemiological applications, focussing on the recent Ebola outbreak in West Africa and a potential emergence of Diphtheria in African refugee camps. Second, I will focus on a macroevolutionary application, shedding light on the radiation of penguins.\n\n
Reconstructing branching lineages in single cell genomics
Freitag, 15.7.16, 12:00-13:00, Raum 404, Eckerstr. 1
Single-cell technologies have recently gained popularity in developmental biology because they allow resolving potential heterogeneities due to asynchronicity of differentiating cells. Popular multivariate approaches for analyzing such data are based on data normalization, followed by dimension reduction and clustering to identify subgroups. However, in the case of cellular differentiation, we cannot expect clear clusters to be present - instead cells tend to follow continuous branching lineages.\n\nWe show that modeling the high-dimensional state space as a diffusion process, where cells move to close-by cells with a distance-dependent probability well reflects the differentiating characteristics. Based on the underlying diffusion map transition kernel, we then propose to order cells according to a diffusion pseudo time, which measures transitions between cells using random walks of arbitrary length. This allows for a robust identification of branching decisions and corresponding trajectories of single cells. We demonstrate the method on single-cell qPCR data of differentiating mouse haematopoietic stem cells as well as on RNA sequencing profiles of embryonic stem cells.\n\nAs outlook if time permits, I will outline how to use this pseudotime in combination with dynamic models to construct a mechanistic understanding of the regulatory process, based on recent work regarding ODE-constrained mixture modeling.