A syringe-sharing model for the spread of HIV: application to Omsk, Western Siberia.

A system of two differential equations is used to model the transmission dynamics of human immunodeficiency virus between 'persons who inject drugs' (PWIDs) and their syringes. Our vector-borne disease model hinges on a metaphorical urn from which PWIDs draw syringes at random which may or may not be infected and may or may not result in one of the two agents becoming infected. The model's parameters are estimated with data mostly from the city of Omsk in Western Siberia. A linear trend in PWID prevalence in Omsk could only be fitted by considering a time-dependent version of the model captured through a secular decrease in the probability that PWIDs decide to share a syringe. A global sensitivity analysis is performed with 14 parameters considered random variables in order to assess their impact on average numbers infected over a 50-year projection. With obvious intervention implications the drug injection rate and the probability of syringe-cleansing are the only parameters whose coefficients of correlations with numbers of infected PWIDs and infected syringes have an absolute value close to or larger than 0.40.

A Leslie matrix model for Sicyopterus lagocephalus in La Réunion: sensitivity, uncertainty and research prioritization.

We propose a Leslie matrix model for the population dynamics of Sicyopterus lagocephalus in La Réunion. In order to capture both the amphidromous and the seasonal natures of the species' life history the model has four stages (sea+three river sites) and is cyclical with a 12 month period. Baseline parameters (age-specific fecundity, spatial dispersion patterns and survival rates) were chosen in such a way that the dominant eigenvalue of the year-on-year projection matrix is 1. Large uncertainties on the parameter values preclude the use of the model for management purpose. A sensitivity/uncertainty analysis sheds light on the parameters that cause much of the output to vary and that are poorly known: the life expectancy in rivers and the mortality both at river mouths and during the drift of larvae to sea. The aim is to help policymakers and researchers prioritize data acquisition efforts. The ultimate goal is a sustainable management of Sicyopterus lagocephalus in La Réunion.

A node pruning algorithm based on a Fourier amplitude sensitivity test method.

In this paper, we propose a new pruning algorithm to obtain the optimal number of hidden units of a single layer of a fully connected neural network (NN). The technique relies on a global sensitivity analysis of model output. The relevance of the hidden nodes is determined by analysing the Fourier decomposition of the variance of the model output. Each hidden unit is assigned a ratio (the fraction of variance which the unit accounts for) that gives their ranking. This quantitative information therefore leads to a suggestion of the most favorable units to eliminate. Experimental results suggest that the method can be seen as an effective tool available to the user in controlling the complexity in NNs.