Multiplicative Forests for Continuous-Time Processes.
Ontology highlight
ABSTRACT:
SUBMITTER: Weiss JC
PROVIDER: S-EPMC4182941 | biostudies-literature | 2012
REPOSITORIES: biostudies-literature
Publications
Multiplicative Forests for Continuous-Time Processes.
Weiss Jeremy C JC Natarajan Sriraam S Page David D
Advances in neural information processing systems 20120101
Learning temporal dependencies between variables over continuous time is an important and challenging task. Continuous-time Bayesian networks effectively model such processes but are limited by the number of conditional intensity matrices, which grows exponentially in the number of parents per variable. We develop a partition-based representation using regression trees and forests whose parameter spaces grow linearly in the number of node splits. Using a multiplicative assumption we show how to ...[more]
Similar Datasets
Project description:Structural failure time models are causal models for estimating the effect of time-varying treatments on a survival outcome. G-estimation and artificial censoring have been proposed for estimating the model parameters in the presence of time-dependent confounding and administrative censoring. However, most existing methods require manually pre-processing data into regularly spaced data, which may invalidate the subsequent causal analysis. Moreover, the computation and inference are challenging due to the nonsmoothness of artificial censoring. We propose a class of continuous-time structural failure time models that respects the continuous-time nature of the underlying data processes. Under a martingale condition of no unmeasured confounding, we show that the model parameters are identifiable from a potentially infinite number of estimating equations. Using the semiparametric efficiency theory, we derive the first semiparametric doubly robust estimators, which are consistent if the model for the treatment process or the failure time model, but not necessarily both, is correctly specified. Moreover, we propose using inverse probability of censoring weighting to deal with dependent censoring. In contrast to artificial censoring, our weighting strategy does not introduce nonsmoothness in estimation and ensures that resampling methods can be used for inference.
| S-EPMC7646189 | biostudies-literature
Project description:Many works in the literature are in favour of the universality of power-laws. However, more recently, a significant amount of research has exposed more subtle details about this subject. In this paper we present two generalisations that aim to solve these possible antagonistic conceptions in a mathematical framework. The first generalisation allows us to show that a vast range of power-laws can be produced through multiplicative process (for example through a single positive feedback mechanism). The second generalisation shows, by solving Lambert's transcendental equations, that the space of solutions where the Pareto distribution is equal to the distribution produced by the power multiplicative transformation (for a number of input distributions) is infinite.
| S-EPMC7350158 | biostudies-literature
Project description:Dropout is a common occurrence in longitudinal studies. Building upon the pattern-mixture modeling approach within the Bayesian paradigm, we propose a general framework of varying-coefficient models for longitudinal data with informative dropout, where measurement times can be irregular and dropout can occur at any point in continuous time (not just at observation times) together with administrative censoring. Specifically, we assume that the longitudinal outcome process depends on the dropout process through its model parameters. The unconditional distribution of the repeated measures is a mixture over the dropout (administrative censoring) time distribution, and the continuous dropout time distribution with administrative censoring is left completely unspecified. We use Markov chain Monte Carlo to sample from the posterior distribution of the repeated measures given the dropout (administrative censoring) times; Bayesian bootstrapping on the observed dropout (administrative censoring) times is carried out to obtain marginal covariate effects. We illustrate the proposed framework using data from a longitudinal study of depression in HIV-infected women; the strategy for sensitivity analysis on unverifiable assumption is also demonstrated.
| S-EPMC2800163 | biostudies-literature
Project description:Branching processes are a class of continuous-time Markov chains (CTMCs) with ubiquitous applications. A general difficulty in statistical inference under partially observed CTMC models arises in computing transition probabilities when the discrete state space is large or uncountable. Classical methods such as matrix exponentiation are infeasible for large or countably infinite state spaces, and sampling-based alternatives are computationally intensive, requiring integration over all possible hidden events. Recent work has successfully applied generating function techniques to computing transition probabilities for linear multi-type branching processes. While these techniques often require significantly fewer computations than matrix exponentiation, they also become prohibitive in applications with large populations. We propose a compressed sensing framework that significantly accelerates the generating function method, decreasing computational cost up to a logarithmic factor by only assuming the probability mass of transitions is sparse. We demonstrate accurate and efficient transition probability computations in branching process models for blood cell formation and evolution of self-replicating transposable elements in bacterial genomes.
| S-EPMC4775097 | biostudies-literature
Project description:The hidden Markov model (HMM) is a framework for time series analysis widely applied to single-molecule experiments. Although initially developed for applications outside the natural sciences, the HMM has traditionally been used to interpret signals generated by physical systems, such as single molecules, evolving in a discrete state space observed at discrete time levels dictated by the data acquisition rate. Within the HMM framework, transitions between states are modeled as occurring at the end of each data acquisition period and are described using transition probabilities. Yet, whereas measurements are often performed at discrete time levels in the natural sciences, physical systems evolve in continuous time according to transition rates. It then follows that the modeling assumptions underlying the HMM are justified if the transition rates of a physical process from state to state are small as compared to the data acquisition rate. In other words, HMMs apply to slow kinetics. The problem is, because the transition rates are unknown in principle, it is unclear, a priori, whether the HMM applies to a particular system. For this reason, we must generalize HMMs for physical systems, such as single molecules, because these switch between discrete states in "continuous time". We do so by exploiting recent mathematical tools developed in the context of inferring Markov jump processes and propose the hidden Markov jump process. We explicitly show in what limit the hidden Markov jump process reduces to the HMM. Resolving the discrete time discrepancy of the HMM has clear implications: we no longer need to assume that processes, such as molecular events, must occur on timescales slower than data acquisition and can learn transition rates even if these are on the same timescale or otherwise exceed data acquisition rates.
| S-EPMC7896036 | biostudies-literature
Project description:SummaryESTIpop is an R package designed to simulate and estimate parameters for continuous-time Markov branching processes with constant or time-dependent rates, a common model for asexually reproducing cell populations. Analytical approaches to parameter estimation quickly become intractable in complex branching processes. In ESTIpop, parameter estimation is based on a likelihood function with respect to a time series of cell counts, approximated by the Central Limit Theorem for multitype branching processes. Additionally, simulation in ESTIpop via approximation can be performed many times faster than exact simulation methods with similar results.Availability and implementationESTIpop is available as an R package on Github (https://github.com/michorlab/estipop).Supplementary informationSupplementary data are available at Bioinformatics online.
| S-EPMC7520045 | biostudies-literature
Project description:Bayesian statistics provides a framework for the integration of dynamic models with incomplete data to enable inference of model parameters and unobserved aspects of the system under study. An important class of dynamic models is discrete state space, continuous-time Markov processes (DCTMPs). Simulated via the Doob-Gillespie algorithm, these have been used to model systems ranging from chemistry to ecology to epidemiology. A new type of proposal, termed 'model-based proposal' (MBP), is developed for the efficient implementation of Bayesian inference in DCTMPs using Markov chain Monte Carlo (MCMC). This new method, which in principle can be applied to any DCTMP, is compared (using simple epidemiological SIS and SIR models as easy to follow exemplars) to a standard MCMC approach and a recently proposed particle MCMC (PMCMC) technique. When measurements are made on a single-state variable (e.g. the number of infected individuals in a population during an epidemic), model-based proposal MCMC (MBP-MCMC) is marginally faster than PMCMC (by a factor of 2-8 for the tests performed), and significantly faster than the standard MCMC scheme (by a factor of 400 at least). However, when model complexity increases and measurements are made on more than one state variable (e.g. simultaneously on the number of infected individuals in spatially separated subpopulations), MBP-MCMC is significantly faster than PMCMC (more than 100-fold for just four subpopulations) and this difference becomes increasingly large.
| S-EPMC4590508 | biostudies-other
Project description:Intense spiking response of a memory-pattern is believed to play a crucial role both in normal learning and pathology, where it can create biased behavior. We recently proposed a novel model for memory amplification where the simultaneous two-fold increase of all excitatory (AMPAR-mediated) and inhibitory (GABAAR-mediated) synapses in a sub-group of cells that constitutes a memory-pattern selectively amplifies this memory. Here we confirm the cellular basis of this model by validating its major predictions in four sets of experiments, and demonstrate its induction via a whole-cell transduction mechanism. Subsequently, using theory and simulations, we show that this whole-cell two-fold increase of all inhibitory and excitatory synapses functions as an instantaneous and multiplicative amplifier of the neurons' spiking. The amplification mechanism acts through multiplication of the net synaptic current, where it scales both the average and the standard deviation of the current. In the excitation-inhibition balance regime, this scaling creates a linear multiplicative amplifier of the cell's spiking response. Moreover, the direct scaling of the synaptic input enables the amplification of the spiking response to be synchronized with rapid changes in synaptic input, and to be independent of previous spiking activity. These traits enable instantaneous real-time amplification during brief elevations of excitatory synaptic input. Furthermore, the multiplicative nature of the amplifier ensures that the net effect of the amplification is large mainly when the synaptic input is mostly excitatory. When induced on all cells that comprise a memory-pattern, these whole-cell modifications enable a substantial instantaneous amplification of the memory-pattern when the memory is activated. The amplification mechanism is induced by CaMKII dependent phosphorylation that doubles the conductance of all GABAA and AMPA receptors in a subset of neurons. This whole-cell transduction mechanism enables both long-term induction of memory amplification when necessary and extinction when not further required.
| S-EPMC5245787 | biostudies-literature
Project description:Forests play a key role in regulating the global carbon cycle, and yet the abiotic and biotic conditions that drive the demographic processes that underpin forest carbon dynamics remain poorly understood in natural ecosystems. To address this knowledge gap, we used repeat forest inventory data from 92,285 trees across four large permanent plots (4-25 ha in size) in temperate mixed forests in northeast China to ask the following questions: (1) How do soil conditions and stand age drive biomass demographic processes? (2) How do vegetation quality (i.e., functional trait diversity and composition) and quantity (i.e., initial biomass stocks) influence biomass demographic processes independently from soil conditions and stand age? (3) What is the relative contribution of growth, recruitment, and mortality to net biomass change? Using structural equation modeling, we showed that all three demographic processes were jointly constrained by multiple abiotic and biotic factors and that mortality was the strongest determinant on net biomass change over time. Growth and mortality, as well as functional trait diversity and the community-weighted mean of specific leaf area (CWMSLA ), declined with stand age. By contrast, high soil phosphorous concentrations were associated with greater functional diversity and faster dynamics (i.e., high growth and mortality rates), but associated with lower CWMSLA and initial biomass stock. More functionally diverse communities also had higher recruitment rates, but did not exhibit faster growth and mortality. Instead, initial biomass stocks and CWMSLA were stronger predictors of biomass growth and mortality, respectively. By integrating the full spectrum of abiotic and biotic drivers of forest biomass dynamics, our study provides critical system-level insights needed to predict the possible consequences of regional changes in forest diversity, composition, structure and function in the context of global change.
| S-EPMC6849813 | biostudies-literature
Project description:While causal mediation analysis has seen considerable recent development for a single measured mediator (M) and final outcome (Y), less attention has been given to repeatedly measured M and Y. Previous methods have typically involved discrete-time models that limit inference to the particular measurement times used and do not recognize the continuous nature of the mediation process over time. To overcome such limitations, we present a new continuous-time approach to causal mediation analysis that uses a differential equations model in a potential outcomes framework to describe the causal relationships among model variables over time. A connection between the differential equation models and standard repeated measures models is made to provide convenient model formulation and fitting. A continuous-time extension of the sequential ignorability assumption allows for identifiable natural direct and indirect effects as functions of time, with estimation based on a two-step approach to model fitting in conjunction with a continuous-time mediation formula. Novel features include a measure of an overall mediation effect based on the "area between the curves," and an approach for predicting the effects of new interventions. Simulation studies show good properties of estimators and the new methodology is applied to data from a cohort study to investigate sugary drink consumption as a mediator of the effect of socioeconomic status on dental caries in children.
| S-EPMC6731141 | biostudies-literature