Dynamic Bayesian Network Analysis Research Articles

Proof of concept for quantitative adverse outcome pathway modeling of chronic toxicity in repeated exposure

Adverse Outcome Pathway (AOP) is a useful tool to glean mode of action (MOE) of a chemical. However, in order to use it for the purpose of risk assessment, an AOP needs to be quantified using in vitro or in vivo data. Majority of quantitative AOPs developed so far, were for single exposure to progressively higher doses. Limited attempts were made to include time in the modeling. Here as a proof-of concept, we developed a hypothetical AOP, and quantified it using a virtual dataset for six repeated exposures using a Bayesian Network Analysis (BN) framework. The virtual data was generated using realistic assumptions. Effects of each exposure were analyzed separately using a static BN model and analyzed in combination using a dynamic BN (DBN) model. Our work shows that the DBN model can be used to calculate the probability of adverse outcome when other upstream KEs were observed earlier. These probabilities can help in identification of early indicators of AO. In addition, we also developed a data driven AOP pruning technique using a lasso-based subset selection, and show that the causal structure of AOP is itself dynamic and changes over time. This proof-of-concept study revealed the possibility for expanding the applicability of the AOP framework to incorporate biological dynamism in toxicity appearance by repeated insults.

Exploring the Pirate Attack Process Risk along the Maritime Silk Road via Dynamic Bayesian Network Analysis

The Maritime Silk Road (MSR) is an important channel for maritime trade between China and other countries in the world. Maritime piracy has brought huge security risks to ships’ navigation and has seriously threatened the lives and property of crew members. To reduce the likelihood of attacks from pirates, it is necessary to study the risk to a ship exposed to attacks from pirates on the MSR. Firstly, risk factors were established from three risk component categories (hazard, mitigation capacity, and vulnerability and exposure) and the risk index system of piracy and armed robbery events was founded. Secondly, the dynamic Bayesian network (DBN) method was introduced to establish a pirate attack risk assessment model ad to conduct a quantitative analysis of the process risk of a ship being attacked by pirates. Finally, combined with the scene data of the MSR, the process risk of a ship being attacked by pirates was modeled and applied as an example. The results showed that the overall risk of a ship being attacked by pirates is the lowest in July and the highest in March. In the whole route, when the ship was in the Gulf of Guinea, the Gulf of Aden–Arabian Sea, and the Strait of Malacca, the risk of pirate attack was the highest. This dynamic network model can effectively analyze the level of risk of pirate attacks on ships, providing a reference for the safety decision-making of ships on ocean routes.

The temporal dependencies between social, emotional and physical health factors in young people receiving mental healthcare: a dynamic Bayesian network analysis.

The needs of young people attending mental healthcare can be complex and often span multiple domains (e.g., social, emotional and physical health factors). These factors often complicate treatment approaches and contribute to poorer outcomes in youth mental health. We aimed to identify how these factors interact over time by modelling the temporal dependencies between these transdiagnostic social, emotional and physical health factors among young people presenting for youth mental healthcare. Dynamic Bayesian networks were used to examine the relationship between mental health factors across multiple domains (social and occupational function, self-harm and suicidality, alcohol and substance use, physical health and psychiatric syndromes) in a longitudinal cohort of 2663 young people accessing youth mental health services. Two networks were developed: (1) 'initial network', that shows the conditional dependencies between factors at first presentation, and a (2) 'transition network', how factors are dependent longitudinally. The 'initial network' identified that childhood disorders tend to precede adolescent depression which itself was associated with three distinct pathways or illness trajectories; (1) anxiety disorder; (2) bipolar disorder, manic-like experiences, circadian disturbances and psychosis-like experiences; (3) self-harm and suicidality to alcohol and substance use or functioning. The 'transition network' identified that over time social and occupational function had the largest effect on self-harm and suicidality, with direct effects on ideation (relative risk [RR], 1.79; CI, 1.59-1.99) and self-harm (RR, 1.32; CI, 1.22-1.41), and an indirect effect on attempts (RR, 2.10; CI, 1.69-2.50). Suicide ideation had a direct effect on future suicide attempts (RR, 4.37; CI, 3.28-5.43) and self-harm (RR, 2.78; CI, 2.55-3.01). Alcohol and substance use, physical health and psychiatric syndromes (e.g., depression and anxiety, at-risk mental states) were independent domains whereby all direct effects remained within each domain over time. This study identified probable temporal dependencies between domains, which has causal interpretations, and therefore can provide insight into their differential role over the course of illness. This work identified social, emotional and physical health factors that may be important early intervention and prevention targets. Improving social and occupational function may be a critical target due to its impacts longitudinally on self-harm and suicidality. The conditional independence of alcohol and substance use supports the need for specific interventions to target these comorbidities.

Risk Response Strategies Selection over the Life Cycle of Project Portfolio

The successful implementation of project portfolios (PP) calls for effective risk management, in which selecting optimal risk response strategies help to reduce the impact of risk. Project portfolio risks (PPRs) exhibit causality and time dependency over the life cycle, which result in cumulative effects over time. By accounting for these risk correlations, risk response could be more effective in reducing expected losses than risk independence assumption. To support effective and sustainable risk management, this study proposes a novel risk response method that integrates the dynamic Bayesian network (DBN) model and reward–risk optimization model to select risk response strategies for different stages of the PP life cycle. The proposed method supports a more comprehensive analysis of risk contagion paths by opening the black box of the risk propagation paths during the PP life cycle. In this method, the PPRs, as the DBN nodes, are first identified, considering the project’s interdependency. Second, DBN analysis is used to assess PPRs by visually modeling the causality and life cycle correlation among risks. Then, the reward–risk optimization model is built to determine risk response strategies for each stage of the life cycle under the constraints. Finally, the proposed method selects risk response strategies for different stages of the PP life cycle. The findings reveal that the risk response effects are maximized if the risks are responded to in earlier stages. Moreover, the findings contribute to helping managers choose the optimal risk response strategies consistent with the risk response budget. As the effect of the strategy depends on the actual situation of the PP, the factors affecting the response effect of the strategies are recommended for further study.

Significance of Hyperhomocysteinemia in Immediate As Well As Long-Term Health Risk/s in Women with Polycystic Ovary Syndrome: a Probabilistic Model Using Dynamic Bayesian Network Analysis.

Polycystic ovary syndrome (PCOS) is a heterogeneous entity comprising broad spectra of ovarian disorders with trademark features of metabolic syndrome like insulin resistance, obesity, and dyslipidaemia to name a few. Hyperhomocysteinemia, an independent risk factor of metabolic syndrome, has been suggested as a causative factor in spontaneous miscarriage in PCOS. However, it is yet to be resolved whether hyperhomocysteinemia has a contributory role in the pathogenesis or could direct long-term sequences of the syndrome. A total of 2355 women with history of one or more first trimester abortions were screened and 1539 were selected for the study. Selected patients were initially divided by the presence or absence of PCOS, while subsequent stratification was based on hyperhomocysteinemia, insulin resistance, and/or obesity. The miscarriage population/s was mostly represented by hyperhomocysteinemia in both the cohorts (PCOS: 69.08% vs. non-PCOS: 56.68%). ROC-AUC values suggest increased predisposition of hyperhomocysteinemia-mediated miscarriage (hyperhomocysteinemia: 0.778; insulin resistance: 0.601; BMI: 0.548). A probabilistic causal model was designed using dynamic Bayesian network to evaluate the time-series data points before, during, and after pregnancy which revealed a possibility of 32.24% (n = 79) of PCOS cohort developing hypertension, 26.94% (n = 66) of onset of diabetes and 4.49% cardiovascular disease 3years following pregnancy. We conclude hyperhomocysteinemia may possibly contribute to spontaneous miscarriage and related to metabolic derailments later in life.

Prediction of visibility in the Arctic based on dynamic Bayesian network analysis

Prediction of visibility in the Arctic based on dynamic Bayesian network analysis

Offshore drilling accident analysis based on Dynamic Bayesian Network

Offshore drilling operations has the characteristics of far offshore, harsh environment, high risk and great technical difficulty. It is difficult to deal with offshore drilling accidents, and it is easy to cause serious social and economic problems. Based on the risk records of drilling engineering, this paper analyzes the forms and causes of offshore drilling accidents, and establishes a dynamic Bayesian network analysis model for offshore drilling accidents. With the dynamic Bayesian network, the prediction probability of drilling risk is calculated. According to the blowout accident in offshore drilling, the analysis results show that the risk of blowout accident is is as high as 50.6%. The research results show that the method and the model presented in the paper is contributed to analysis and prevention of offshore drilling accidents.

Dynamic Bayesian Network Analysis of Inflammation Biomarkers Reveals Age-Related Changes of the Inflammatory Response after Trauma

Dynamic Bayesian Network Analysis of Inflammation Biomarkers Reveals Age-Related Changes of the Inflammatory Response after Trauma

Reliability theory based on quality risk network analysis for Chinese medicine injection

A new risk analysis method based upon reliability theory was introduced in this paper for the quality risk management of Chinese medicine injection manufacturing plants. The risk events including both cause and effect ones were derived in the framework as nodes with a Bayesian network analysis approach. It thus transforms the risk analysis results from failure mode and effect analysis (FMEA) into a Bayesian network platform. With its structure and parameters determined, the network can be used to evaluate the system reliability quantitatively with probabilistic analytical appraoches. Using network analysis tools such as GeNie and AgenaRisk, we are able to find the nodes that are most critical to influence the system reliability. The importance of each node to the system can be quantitatively evaluated by calculating the effect of the node on the overall risk, and minimization plan can be determined accordingly to reduce their influences and improve the system reliability. Using the Shengmai injection manufacturing plant of SZYY Ltd as a user case, we analyzed the quality risk with both static FMEA analysis and dynamic Bayesian Network analysis. The potential risk factors for the quality of Shengmai injection manufacturing were identified with the network analysis platform. Quality assurance actions were further defined to reduce the risk and improve the product quality.

Approaching dynamic reliability with predictive and diagnostic purposes by exploiting dynamic Bayesian networks

We talk about dynamic reliability when the reliability parameters of the system, such as the failure rates, vary according to the current state of the system. In this article, several versions of a benchmark on dynamic reliability taken from the literature are examined. Each version deals with particular aspects such as state-dependent failure rates, failure on demand, and repair. In dynamic reliability evaluation, the complete behavior of the system has to be taken into account, instead of the only failure propagation as in fault tree analysis. To this aim, we exploit dynamic Bayesian networks and the software tool RADYBAN ( Reliability Analysis with DYnamic BAyesian Networks), with the goal of computing the system unreliability. Because of the coherence between the results returned by dynamic Bayesian network analysis and those obtained by means of other methods, together with the possibility to compute diagnostic indices, we propose dynamic Bayesian network and RADYBAN to be a valid approach to dynamic reliability evaluation.

Combined network and mathematical modeling suggests novel inflammation positive feedback circuit and role for IL-1α in experimental trauma/hemorrhage and hepatocyte hypoxia

13 Combined network and mathematical modeling suggests novel inflammation positive feedback circuit and role for IL-1 in experimental trauma/hemorrhage and hepatocyte hypoxia Nabil Azhar, Rami Namas, Bahiyyah Jefferson, Derek Barclay, Ruben Zamora, Yoram Vodovotz University of Pittsburgh, Pittsburgh, PA, USA Objectives: Trauma/hemorrhagic shock (T/HS) is the most common cause of death for young people in the United States, costing more than $400 billion annually. Most deaths from T/HS occur due to the multiple organ dysfunction syndrome. Multiple organ dysfunction syndrome is thought to be due, in part, to dysregulated inflammation. We have used computational modeling to define key networks and mechanisms of post-T/HS acute inflammation. We used a Dynamic Bayesian Network (DBN) algorithm (adapted from Grzegorczyk and Husmeier, Bioinformatics 2010;27:693). Given time-series data, DBNs provide a way of inferring causal relationships based on probabilistic measures. This approximation helps to suggest the overall network structure, including central nodes that exhibit feedback. Methods: C57Bl/6 mice were subjected to T/HS for 0 to 4 hours. Separately, C57Bl/6 hepatocytes were subjected to an in vitro surrogate of T/HS, namely, hypoxia (culture at 1% O2 for 1 to 72 hours). Cytokines/chemokines were assayed in the plasma (in vivo T/HS) as well as supernatants and lysates (in vitro hypoxia) by Luminex (MiraiBio, Alameda, CA), and the data were subjected to DBN analysis. Results: Dynamic Bayesian Network analysis suggested a core network that involves the cytokines IL-10 and IL-1α, as well as the chemokines IP-10 and MIG. Assigning biologically plausible functions to directed edges of the network resulted in a so-called incoherent type I feed forward loop that predicted pulsatile, “gatekeeper” behavior for the cytokine IL-1α—a behavior that was indeed observed in preliminary studies in mice subjected to T/ HS. An initial ordinary differential equation (ODE) model was constructed to analyze this behavior in further detail. Simulated trajectories of cytokines and chemokines in the ODE model corresponded closely with the data. Conclusions: Data-driven (DBN) and mechanistic (ODE) analyses suggest a novel inflammatory circuit with a central role for IL-1α in T/HS and hepatocyte hypoxia. http://dx.doi.org/10.1016/j.jcrc.2012.10.029

Possible novel roles of poly(rC)-binding protein 1 in SH-SY5Y neurocytes: an analysis using a dynamic Bayesian network

Poly(rC)-binding protein 1 (PCBP1) belongs to the heterogeneous nuclear ribonucleoprotein family and participates in transcriptional and translational regulation. Previous work has identified transcripts targeted by both knockdown and overexpression of PCBP1 in SH-SY5Y neuroblastoma cells using a microarray or ProteomeLab protein fractionation 2-dimensions (PF-2D) and quadrupole time-of-flight mass spectrometer. The present study aimed to further determine whether these altered transcripts from major pathways (such as Wnt signaling, TGF-β signaling, cell cycling, and apoptosis) and two other genes, H2AFX and H2BFS (screened by PF-2D), have spatial relationships. The genes were studied by qRT-PCR, and dynamic Bayesian network analysis was used to rebuild the coordination network of these transcripts. PCBP1 controlled the expression or activity of the seven transcripts. Moreover, PCBP1 indirectly regulated MAP2K2, FOS, FST, TP53 and WNT7B through H2AFX or regulated these genes through SAT. In contrast, TP53 and WNT7B are regulated by other genes. The seven transcripts and PCBP1 are closely associated in a spatial interaction network.

Large-scale dynamic gene regulatory network inference combining differential equation models with local dynamic Bayesian network analysis

Reverse engineering gene regulatory networks, especially large size networks from time series gene expression data, remain a challenge to the systems biology community. In this article, a new hybrid algorithm integrating ordinary differential equation models with dynamic Bayesian network analysis, called Differential Equation-based Local Dynamic Bayesian Network (DELDBN), was proposed and implemented for gene regulatory network inference. The performance of DELDBN was benchmarked with an in vivo dataset from yeast. DELDBN significantly improved the accuracy and sensitivity of network inference compared with other approaches. The local causal discovery algorithm implemented in DELDBN also reduced the complexity of the network inference algorithm and improved its scalability to infer larger networks. We have demonstrated the applicability of the approach to a network containing thousands of genes with a dataset from human HeLa cell time series experiments. The local network around BRCA1 was particularly investigated and validated with independent published studies. BRAC1 network was significantly enriched with the known BRCA1-relevant interactions, indicating that DELDBN can effectively infer large size gene regulatory network from time series data. The R scripts are provided in File 3 in Supplementary Material. zheng.li@monsanto.com; jingdong.liu@monsanto.com Supplementary data are available at Bioinformatics online.

A cell-type-specific dynamic Bayesian network model for spontaneous and optogenetically evoked activity in the primary visual cortex

Reciprocal interaction between excitatory and inhibitory neurons within and between layers of the cerebral cortex is a major element of brain function. Ensemble extracellular recording techniques using mircroelectrode arrays have permitted recording spiking activity of many neurons simultaneously to characterize network function [1]. Identifying the type of neurons in these recordings is not straightforward due to the variability in extracellular spike shapes, and the irregularities often observed in their interspike interval characteristics. In this study, we used optogenetic tools to genetically target fast spiking interneurons in the primary visual cortex of mice [2]. We modulated their spiking activity by illuminating the region with very short pulses (<1 ms) of light (~470 nm wavelength)in mice primary visual cortex. Using Dynamic Bayesian Network analysis [3], we assessed the effective connectivity between the recorded neurons in the presence and absence of light stimuli under distinct cortical states observed under light anesthesia. DBNs could identify the effective connectivity between putative excitatory pyramidal cells and inhibitory interneurons. These findings suggest a novel and unprecedented way to identify cortical neuronal circuits and characterize the dynamics of their computations in vivo.

Proinflammatory Caspase-2-Mediated Macrophage Cell Death Induced by a Rough Attenuated Brucella suis Strain

Brucella spp. are intracellular bacteria that cause an infectious disease called brucellosis in humans and many domestic and wildlife animals. B. suis primarily infects pigs and is pathogenic to humans. The macrophage-Brucella interaction is critical for the establishment of a chronic Brucella infection. Our studies showed that smooth virulent B. suis strain 1330 (S1330) prevented programmed cell death of infected macrophages and rough attenuated B. suis strain VTRS1 (a vaccine candidate) induced strong macrophage cell death. To further investigate the mechanism of VTRS1-induced macrophage cell death, microarrays were used to analyze temporal transcriptional responses of murine macrophage-like J774.A1 cells infected with S1330 or VTRS1. In total 17,685 probe sets were significantly regulated based on the effects of strain, time and their interactions. A miniTUBA dynamic Bayesian network analysis predicted that VTRS1-induced macrophage cell death was mediated by a proinflammatory gene (the tumor necrosis factor alpha [TNF-α] gene), an NF-κB pathway gene (the IκB-α gene), the caspase-2 gene, and several other genes. VTRS1 induced significantly higher levels of transcription of 40 proinflammatory genes than S1330. A Mann-Whitney U test confirmed the proinflammatory response in VTRS1-infected macrophages. Increased production of TNF-α and interleukin 1β (IL-1β) were also detected in the supernatants in VTRS1-infected macrophage cell culture. Hyperphosphorylation of IκB-α was observed in macrophages infected with VTRS1 but not S1330. The important roles of TNF-α and IκB-α in VTRS1-induced macrophage cell death were further confirmed by individual inhibition studies. VTRS1-induced macrophage cell death was significantly inhibited by a caspase-2 inhibitor but not a caspase-1 inhibitor. The role of caspase-2 in regulating the programmed cell death of VTRS1-infected macrophages was confirmed in another study using caspase-2-knockout mice. In summary, VTRS1 induces a proinflammatory, caspase-2- and NF-κB-mediated macrophage cell death. This unique cell death differs from apoptosis, which is not proinflammatory. It is also different from classical pyroptosis, which is caspase-1 mediated.

Hybrid ICA–Bayesian network approach reveals distinct effective connectivity differences in schizophrenia

We utilized a discrete dynamic Bayesian network (dDBN) approach (Burge, J., Lane, T., Link, H., Qiu, S., Clark, V.P., 2007. Discrete dynamic Bayesian network analysis of fMRI data. Hum Brain Mapp.) to determine differences in brain regions between patients with schizophrenia and healthy controls on a measure of effective connectivity, termed the approximate conditional likelihood score (ACL) (Burge, J., Lane, T., 2005. Learning Class-Discriminative Dynamic Bayesian Networks. Proceedings of the International Conference on Machine Learning, Bonn, Germany, pp. 97-104.). The ACL score represents a class-discriminative measure of effective connectivity by measuring the relative likelihood of the correlation between brain regions in one group versus another. The algorithm is capable of finding non-linear relationships between brain regions because it uses discrete rather than continuous values and attempts to model temporal relationships with a first-order Markov and stationary assumption constraint (Papoulis, A., 1991. Probability, random variables, and stochastic processes. McGraw-Hill, New York.). Since Bayesian networks are overly sensitive to noisy data, we introduced an independent component analysis (ICA) filtering approach that attempted to reduce the noise found in fMRI data by unmixing the raw datasets into a set of independent spatial component maps. Components that represented noise were removed and the remaining components reconstructed into the dimensions of the original fMRI datasets. We applied the dDBN algorithm to a group of 35 patients with schizophrenia and 35 matched healthy controls using an ICA filtered and unfiltered approach. We determined that filtering the data significantly improved the magnitude of the ACL score. Patients showed the greatest ACL scores in several regions, most markedly the cerebellar vermis and hemispheres. Our findings suggest that schizophrenia patients exhibit weaker connectivity than healthy controls in multiple regions, including bilateral temporal, frontal, and cerebellar regions during an auditory paradigm.

Discrete dynamic Bayesian network analysis of fMRI data.

We examine the efficacy of using discrete Dynamic Bayesian Networks (dDBNs), a data-driven modeling technique employed in machine learning, to identify functional correlations among neuroanatomical regions of interest. Unlike many neuroimaging analysis techniques, this method is not limited by linear and/or Gaussian noise assumptions. It achieves this by modeling the time series of neuroanatomical regions as discrete, as opposed to continuous, random variables with multinomial distributions. We demonstrated this method using an fMRI dataset collected from healthy and demented elderly subjects (Buckner, et al., 2000: J Cogn Neurosci 12:24-34) and identify correlates based on a diagnosis of dementia. The results are validated in three ways. First, the elicited correlates are shown to be robust over leave-one-out cross-validation and, via a Fourier bootstrapping method, that they were not likely due to random chance. Second, the dDBNs identified correlates that would be expected given the experimental paradigm. Third, the dDBN's ability to predict dementia is competitive with two commonly employed machine-learning classifiers: the support vector machine and the Gaussian naive Bayesian network. We also verify that the dDBN selects correlates based on non-linear criteria. Finally, we provide a brief analysis of the correlates elicited from Buckner et al.'s data that suggests that demented elderly subjects have reduced involvement of entorhinal and occipital cortex and greater involvement of the parietal lobe and amygdala in brain activity compared with healthy elderly (as measured via functional correlations among BOLD measurements). Limitations and extensions to the dDBN method are discussed.

MiniTUBA: medical inference by network integration of temporal data using Bayesian analysis

Many biomedical and clinical research problems involve discovering causal relationships between observations gathered from temporal events. Dynamic Bayesian networks are a powerful modeling approach to describe causal or apparently causal relationships, and support complex medical inference, such as future response prediction, automated learning, and rational decision making. Although many engines exist for creating Bayesian networks, most require a local installation and significant data manipulation to be practical for a general biologist or clinician. No software pipeline currently exists for interpretation and inference of dynamic Bayesian networks learned from biomedical and clinical data. miniTUBA is a web-based modeling system that allows clinical and biomedical researchers to perform complex medical/clinical inference and prediction using dynamic Bayesian network analysis with temporal datasets. The software allows users to choose different analysis parameters (e.g. Markov lags and prior topology), and continuously update their data and refine their results. miniTUBA can make temporal predictions to suggest interventions based on an automated learning process pipeline using all data provided. Preliminary tests using synthetic data and laboratory research data indicate that miniTUBA accurately identifies regulatory network structures from temporal data. miniTUBA is available at http://www.minituba.org.