Bayesian Markov Chain Monte Carlo Research Articles

Bayesian Estimation of Finite-Horizon Discrete Choice Dynamic Programming Models

We develop a Bayesian Markov chain Monte Carlo (MCMC) algorithm for estimating finite-horizon discrete choice dynamic programming (DDP) models. The proposed algorithm has the potential to reduce the computational burden significantly when some of the state variables are continuous. In a conventional approach to estimating such a finite-horizon DDP model, researchers achieve a reduction in estimation time by evaluating value functions at only a subset of state points and applying an interpolation method to approximate value functions at the remaining state points (e.g., Keane and Wolpin 1994). Although this approach has proven to be effective, the computational burden could still be high if the model has multiple continuous state variables or the number of periods in the time horizon is large. We propose a new estimation algorithm to reduce the computational burden for estimating this class of models. It extends the Bayesian MCMC algorithm for stationary infinite-horizon DDP models proposed by Imai, Jain and Ching (2009) (IJC). In our algorithm, we solve value functions at only one randomly chosen state point per time period, store those partially solved value functions period by period, and approximate expected value functions nonparametrically using the set of those partially solved value functions. We conduct Monte Carlo exercises and show that our algorithm is able to recover the true parameter values well. Finally, similar to IJC, our algorithm allows researchers to incorporate flexible unobserved heterogeneity, which is often computationally infeasible in the conventional two-step estimation approach (e.g., Hotz and Miller 1993; Aguirregabiria and Mira 2002).

Modeling of Annual Maximum Storm Intensity with Bayesian Markov Chain Monte Carlo (MCMC) and L-moment

This study presents the best fitting distribution to describe the siries MSI based on hourly rainfall form 1970 to 2008 for three rain gauge stations in Peninsular Malaysia namely Bertam, Dungun and Pekan. Two three-parameter extreme value distributions which are considered are Generalized Extreme Value (GEV) and Generalized Logistic (GL). The parameters of these distributions are determined using the Bayesian MCMC with non-informative prior distribution and L-moments (LMOM) method. The Goodness-Of-Fit (GOF) between empirical data and theoretical distributions are then evaluated for each stations. The result show that the majority of the stations are found that the L-moment method can give the best modelling for MSI, specified for GEV distribution. Based on the model that has been identified, we can reasonably predict the risks associated the MSI for various return periods.

Spot modelling of periodic weak-line T Tauri stars observed by CoRoT in NGC 2264

The space telescope CoRoT has provided light curves of T Tauri stars belonging to the star-forming region of NGC 2264 with unprecedented continuity and precision in the framework of a coordinated multi-wavelength observational project. We perform spot modelling of the optical light curves of five weak-line T Tauri stars whose variability is dominated by starspots. We apply two-spot and evolving single spot models in the framework of a Bayesian Monte Carlo Markov Chain approach to derive the a posteriori distribution of the starspot parameters and the inclination of the star rotation axis. We focus on the rotation periods of the spots that can provide evidence for differential rotation in those stars. We find meaningful results in the case of three stars with an inclination higher than 50 deg and a slow variation of the light curve amplitude. The relative difference of the spot rotation periods ranges from 0.02 to 0.05 that is 3-10 times larger than the amplitude of the differential rotation found in similar stars with Doppler imaging techniques. We conclude that the intrinsic starspot evolution, although very slow, has a significant impact on the determination of the differential rotation by means of our spot modelling approach. We estimate typical timescales for the evolution of the starspot pattern between about 20 and 50 rotation periods in our stars.

Estimating the Effective Sample Size of Tree Topologies from Bayesian Phylogenetic Analyses.

Bayesian phylogenetic analyses estimate posterior distributions of phylogenetic tree topologies and other parameters using Markov chain Monte Carlo (MCMC) methods. Before making inferences from these distributions, it is important to assess their adequacy. To this end, the effective sample size (ESS) estimates how many truly independent samples of a given parameter the output of the MCMC represents. The ESS of a parameter is frequently much lower than the number of samples taken from the MCMC because sequential samples from the chain can be non-independent due to autocorrelation. Typically, phylogeneticists use a rule of thumb that the ESS of all parameters should be greater than 200. However, we have no method to calculate an ESS of tree topology samples, despite the fact that the tree topology is often the parameter of primary interest and is almost always central to the estimation of other parameters. That is, we lack a method to determine whether we have adequately sampled one of the most important parameters in our analyses. In this study, we address this problem by developing methods to estimate the ESS for tree topologies. We combine these methods with two new diagnostic plots for assessing posterior samples of tree topologies, and compare their performance on simulated and empirical data sets. Combined, the methods we present provide new ways to assess the mixing and convergence of phylogenetic tree topologies in Bayesian MCMC analyses.

An Application of Bayesian Approach in Modeling Risk of Death in an Intensive Care Unit.

Background and ObjectivesThere are not many studies that attempt to model intensive care unit (ICU) risk of death in developing countries, especially in South East Asia. The aim of this study was to propose and describe application of a Bayesian approach in modeling in-ICU deaths in a Malaysian ICU.MethodsThis was a prospective study in a mixed medical-surgery ICU in a multidisciplinary tertiary referral hospital in Malaysia. Data collection included variables that were defined in Acute Physiology and Chronic Health Evaluation IV (APACHE IV) model. Bayesian Markov Chain Monte Carlo (MCMC) simulation approach was applied in the development of four multivariate logistic regression predictive models for the ICU, where the main outcome measure was in-ICU mortality risk. The performance of the models were assessed through overall model fit, discrimination and calibration measures. Results from the Bayesian models were also compared against results obtained using frequentist maximum likelihood method.ResultsThe study involved 1,286 consecutive ICU admissions between January 1, 2009 and June 30, 2010, of which 1,111 met the inclusion criteria. Patients who were admitted to the ICU were generally younger, predominantly male, with low co-morbidity load and mostly under mechanical ventilation. The overall in-ICU mortality rate was 18.5% and the overall mean Acute Physiology Score (APS) was 68.5. All four models exhibited good discrimination, with area under receiver operating characteristic curve (AUC) values approximately 0.8. Calibration was acceptable (Hosmer-Lemeshow p-values > 0.05) for all models, except for model M3. Model M1 was identified as the model with the best overall performance in this study.ConclusionFour prediction models were proposed, where the best model was chosen based on its overall performance in this study. This study has also demonstrated the promising potential of the Bayesian MCMC approach as an alternative in the analysis and modeling of in-ICU mortality outcomes.

Phylodynamics of avian influenza clade 2.2.1 H5N1 viruses in Egypt.

BackgroundHighly pathogenic avian influenza (HPAI) viruses of the H5N1 subtype are widely distributed within poultry populations in Egypt and have caused multiple human infections. Linking the epidemiological and sequence data is important to understand the transmission, persistence and evolution of the virus. This work describes the phylogenetic dynamics of H5N1 based on molecular characterization of the hemagglutinin (HA) gene of isolates collected from February 2006 to May 2014.MethodsFull-length HA sequences of 368 H5N1 viruses were generated and were genetically analysed to study their genetic evolution. They were collected from different poultry species, production sectors, and geographic locations in Egypt. The Bayesian Markov Chain Monte Carlo (BMCMC) method was applied to estimate the evolutionary rates among different virus clusters; additionally, an analysis of selection pressures in the HA gene was performed using the Single Likelihood Ancestor Counting (SLAC) method.ResultsThe phylogenetic analysis of the H5 gene from 2006–14 indicated the presence of one virus introduction of the classic clade (2.2.1) from which two main subgroups were originated, the variant subgroup which was further subdivided into 2 sub-divisions (2.2.1.1 and 2.2.1.1a) and the endemic subgroup (2.2.1.2). The clade 2.2.1.2 showed a high evolution rate over a period of 6 years (6.9 × 10−3 sub/site/year) in comparison to the 2.2.1.1a variant cluster (7.2 × 10−3 over a period of 4 years). Those two clusters are under positive selection as they possess 5 distinct positively selected sites in the HA gene. The mutations at 120, 154, and 162 HA antigenic sites and the other two mutations (129∆, I151T) that occurred from 2009–14 were found to be stable in the 2.2.1.2 clade. Additionally, 13 groups of H5N1 HPAI viruses were identified based on their amino acid sequences at the cleavage site and “EKRRKKR” became the dominant pattern beginning in 2013.ConclusionsContinuous evolution of H5N1 HPAI viruses in Egypt has been observed in all poultry farming and production systems in almost all regions of the country. The wide circulation of the 2.2.1.2 clade carrying triple mutations (120, 129∆, I151T) associated with increased binding affinity to human receptors is an alarming finding of public health importance.Electronic supplementary materialThe online version of this article (doi:10.1186/s12985-016-0477-7) contains supplementary material, which is available to authorized users.

Reducing uncertainty in flood frequency analyses: A comparison of local and regional approaches involving information on extreme historical floods

Summary This paper proposes a detailed comparison of local and regional approaches for flood frequency analyses, with a special emphasis on the effects of (a) the information on extreme floods used in the analysis (historical data or recent extreme floods observed at ungauged sites), and (b) the assumptions associated with regional approaches (statistical homogeneity of considered series, independence of observations). The results presented are based on two case studies: the Ard e che and Argens rivers regions in south-east of France. Four approaches are compared: 1 – local analysis based on continuous measured series, 2 – local analysis with historical information, 3 – regional index-flood analysis based on continuous series, 4 – regional analysis involving information on extremes (including both historical floods and recent floods observed at ungauged sites). The inference approach used is based on a GEV distribution and a Bayesian Monte Carlo Markov Chain approach for parameters estimation. The comparison relies both on (1) available observed datasets and (2) Monte Carlo simulations in order to evaluate the effects of sampling variability and to analyze the possible influence of regional heterogeneities. The results indicate that a relatively limited level of regional heterogeneity, which may not be detected through homogeneity tests, may significantly affect the performances of regional approaches. These results also illustrate the added value of information on extreme floods, historical floods or recent floods observed at ungauged sites, in both local and regional approaches. As far as possible, gathering such information and incorporating it into flood frequency studies should be promoted. Finally, the presented Monte Carlo simulations appear as an interesting analysis tool for adapting the estimation strategy to the available data for each specific case study.

Bayesian Inference, Stochastic Simulation and Their Applications in Wireless Communication Systems

Bayesian inference has been successfully applied in fields as varied as anti-spam filtering, DNA sequencing, war codebreaking and election forecasting. Founded on the apparently simple Bayes’ theorem, which relates the previous distribution of a parameter with its distribution after evidence is collected, Bayesian tools allow for incorporating all existing knowledge about the phenomenon under study in order to improve parameter estimation. Because of the stochastic nature of the wireless channel, Bayesian inference is particularly well suited to the problem of symbol detection in many modern digital communication systems. When combined with Markov Chain Monte Carlo (MCMC) techniques, Bayesian receivers are capable of achieving minimum Bit Error Rate (BER) while avoiding the prohibitively high computational complexity associated with standard Maximum Likelihood (ML) or Maximum A Posteriori (MAP) estimators. In addition, such receivers are capable of numerically integrating out channel coefficients and noise variance, thus avoiding the need to use sub-optimal estimates of these parameters. This tutorial presents the rudiments of Bayesian statistics and MCMC in general, and discusses their applications in wireless communications in particular. The paper also details the design of Bayesian MCMC receiver in a system employing BPSK and ubject frequency-selective fading and Gaussian noise. Afterwards, recent advances in Bayesian receivers are surveyed for several important practical wireless transmission schemes, including MIMO, CDMA and OFDM. In addition, the paper addresses the application of Bayesian tools in challenging channel conditions — namely, nonlinear, non-Gaussian, underwater and fast fading channels.

Tourism sector, Travel agencies, and Transport Suppliers: Comparison of Different Estimators in the Structural Equation Modeling

Abstract The paper addresses the effect of external integration (EI) with transport suppliers on the efficiency of travel agencies in the tourism sector supply chains. The main aim is the comparison of different estimation methods used in the structural equation modeling (SEM), applied to discover possible relationships between EIs and efficiencies. The latter are calculated by the means of data envelopment analysis (DEA). While designing the structural equation model, the exploratory and confirmatory factor analyses are also used as preliminary statistical procedures. For the estimation of parameters of SEM model, three different methods are explained, analyzed and compared: maximum likelihood (ML) method, Bayesian Markov Chain Monte Carlo (BMCMC) method, and unweighted least squares (ULS) method. The study reveals that all estimation methods calculate comparable estimated parameters. The results also give an evidence of good model fit performance. Besides, the research confirms that the amplified external integration with transport providers leads to increased efficiency of travel agencies, which might be a very interesting finding for the operational management.

Comparison of estimation methods for creating small area rates of acute myocardial infarction among Medicare beneficiaries in California

Creating local population health measures from administrative data would be useful for health policy and public health monitoring purposes. While a wide range of options – from simple spatial smoothers to model-based methods – for estimating such rates exists, there are relatively few side-by-side comparisons, especially not with real-world data. In this paper, we compare methods for creating local estimates of acute myocardial infarction rates from Medicare claims data. A Bayesian Monte Carlo Markov Chain estimator that incorporated spatial and local random effects performed best, followed by a method-of-moments spatial Empirical Bayes estimator. As the former is more complicated and time-consuming, spatial linear Empirical Bayes methods may represent a good alternative for non-specialist investigators.

Equilibrium model constraints on baryon cycling across cosmic time

Galaxies strongly self-regulate their growth via energetic feedback from stars, supernovae, and black holes, but these processes are among the least understood aspects of galaxy formation theory. We present an analytic galaxy evolution model that directly constrains such feedback processes from observed galaxy scaling relations. The equilibrium model, which is broadly valid for star-forming central galaxies that dominate cosmic star formation, is based on the ansatz that galaxies live in a slowly-evolving equilibrium between inflows, outflows, and star formation. Using a Bayesian Monte Carlo Markov chain approach, we constrain our model to match observed galaxy scaling relations between stellar mass and halo mass, star formation rate, and metallicity from 0<z<2. A good fit (chi^2~1.6) is achieved with eight free parameters. We further show that constraining our model to any two of the three data sets also produces a fit to the third that is within reasonable systematic uncertainties. The resulting best-fit parameters that describe baryon cycling suggest galactic outflow scalings intermediate between energy and momentum-driven winds, a weak dependence of wind recycling time on mass, and a quenching mass scale that evolves modestly upwards with redshift. This model further predicts a stellar mass-star formation rate relation that is in good agreement with observations to z~6. Our results suggest that this simple analytic framework captures the basic physical processes required to model the mean evolution of stars and metals in galaxies, despite not incorporating many canonical ingredients of galaxy formation models such as merging or disk formation.

Modeling two strains of disease via aggregate-level infectivity curves.

Well formulated models of disease spread, and efficient methods to fit them to observed data, are powerful tools for aiding the surveillance and control of infectious diseases. Our project considers the problem of the simultaneous spread of two related strains of disease in a context where spatial location is the key driver of disease spread. We start our modeling work with the individual level models (ILMs) of disease transmission, and extend these models to accommodate the competing spread of the pathogens in a two-tier hierarchical population (whose levels we refer to as 'farm' and 'animal'). The postulated interference mechanism between the two strains is a period of cross-immunity following infection. We also present a framework for speeding up the computationally intensive process of fitting the ILM to data, typically done using Markov chain Monte Carlo (MCMC) in a Bayesian framework, by turning the inference into a two-stage process. First, we approximate the number of animals infected on a farm over time by infectivity curves. These curves are fit to data sampled from farms, using maximum likelihood estimation, then, conditional on the fitted curves, Bayesian MCMC inference proceeds for the remaining parameters. Finally, we use posterior predictive distributions of salient epidemic summary statistics, in order to assess the model fitted.

Transmission and Demographic Dynamics of Coxsackievirus B1.

The infectious activity of coxsackievirus B1 (CV-B1) in Taiwan was high from 2008 to 2010, following an alarming increase in severe neonate disease in the United States (US). To examine the relationship between CV-B1 strains isolated in Taiwan and those from other parts of the world, we performed a phylodynamic study using VP1 and partial 3Dpol (414 nt) sequences from 22 strains of CV-B1 isolated in Taiwan (1989–2010) and compared them to sequences from strains isolated worldwide. Phylogenetic trees were constructed by neighbor-joining, maximum likelihood, and Bayesian Monte Carlo Markov Chain methods. Four genotypes (GI–IV) in the VP1 region of CV-B1 and three genotypes (GA–C) in the 3Dpol region of enterovirus B were identified and had high support values. The phylogenetic analysis indicates that the GI and GIII strains in VP1 were geographically distributed in Taiwan (1993–1994) and in India (2007–2009). On the other hand, the GII and GIV strains appear to have a wider spatiotemporal distribution and ladder-like topology A stair-like phylogeny was observed in the VP1 region indicating that the phylogeny of the virus may be affected by different selection pressures in the specified regions. Further, most of the GI and GII strains in the VP1 tree were clustered together in GA in the 3D tree, while the GIV strains diverged into GB and GC. Taken together, these data provide important insights into the population dynamics of CV-B1 and indicate that incongruencies in specific gene regions may contribute to spatiotemporal patterns of epidemicity for this virus.

Towards a formal genealogical classification of the Lezgian languages (North Caucasus): testing various phylogenetic methods on lexical data.

A lexicostatistical classification is proposed for 20 languages and dialects of the Lezgian group of the North Caucasian family, based on meticulously compiled 110-item wordlists, published as part of the Global Lexicostatistical Database project. The lexical data have been subsequently analyzed with the aid of the principal phylogenetic methods, both distance-based and character-based: Starling neighbor joining (StarlingNJ), Neighbor joining (NJ), Unweighted pair group method with arithmetic mean (UPGMA), Bayesian Markov chain Monte Carlo (MCMC), Unweighted maximum parsimony (UMP). Cognation indexes within the input matrix were marked by two different algorithms: traditional etymological approach and phonetic similarity, i.e., the automatic method of consonant classes (Levenshtein distances). Due to certain reasons (first of all, high lexicographic quality of the wordlists and a consensus about the Lezgian phylogeny among Caucasologists), the Lezgian database is a perfect testing area for appraisal of phylogenetic methods. For the etymology-based input matrix, all the phylogenetic methods, with the possible exception of UMP, have yielded trees that are sufficiently compatible with each other to generate a consensus phylogenetic tree of the Lezgian lects. The obtained consensus tree agrees with the traditional expert classification as well as some of the previously proposed formal classifications of this linguistic group. Contrary to theoretical expectations, the UMP method has suggested the least plausible tree of all. In the case of the phonetic similarity-based input matrix, the distance-based methods (StarlingNJ, NJ, UPGMA) have produced the trees that are rather close to the consensus etymology-based tree and the traditional expert classification, whereas the character-based methods (Bayesian MCMC, UMP) have yielded less likely topologies.

Probabilistic analysis of phytoplankton biomass at the Frisian Inlet (NL)

The BLOOM II model is applied in this study to interrogate the variation of phytoplankton biomass (in terms of chlorophyll a) at the Frisian Inlet (NL), located in the north of the Netherlands. The validation results of the model show that more than 90% of the comparisons between model results and observations have a good agreement, while only 3.1% are classified as poor. Chlorophyll a is significantly correlated with the ecological indicators of Si and NH4 by observational analysis. The modelled chlorophyll a is fitted well by a normal distribution function (μ = 3.95 mg m−3, σ = 2.96 mg m−3). Relationships between the modelled ecological indicators are investigated. Secchi depth as another important indicator is closely linked with chlorophyll a, PAR and Kd. Moreover, particular attention is paid to the phytoplankton biomass in response to nutrient availability. The decrease rate of chlorophyll a is from 16.8% to 19.2% for N-reduction, 17.4%–21.3% for P-reduction and 21.7%–28.0% for both N- and P-reduction. In the presence of uncertainty, the improved prediction of chlorophyll a is derived by Bayesian Markov Chain Monte Carlo (BMCMC) in this study. The improved prediction of chlorophyll a concentrations vary from 0.147 to 15.34 mg m−3 within the 95% confidence interval. This study emphasises the use of an ecological model to predict the variation of phytoplankton biomass, and improves the prediction with the integration of uncertainty analysis.

Bayesian Markov chain Monte Carlo imputation for the transiting exoplanets with an application in clustering analysis

To impute the missing values of mass in the transiting exoplanet data, this paper uses the Frank copula to combine two Pareto marginal distributions. Next, a Bayesian Markov chain Monte Carlo (MCMC) imputation method is proposed. The proposed Bayesian MCMC imputation method is found to outperform the mean imputation method. Clustering analysis can shed light on the formation and evolution of exoplanets. After imputing the missing values of mass in the transiting exoplanet data using the proposed approach, the similarity-based clustering method (SCM) clustering algorithm is applied to the logarithm of mass and period for this complete data set. The SCM clustering result indicates two clusters. Furthermore, the intracluster Spearman rank-order correlation coefficients for mass and period in these two clusters are 0.401 and , respectively, at a significance level of 0.01. This result illustrates that the mass and period correlate in an opposite way between the two different clusters. It implies that the formation and evolution processes of these two clusters are different.

Supplier selection for sustainable operations: A triple-bottom-line approach using a Bayesian framework

In evaluating and selecting sustainable suppliers, we take a triple-bottom-line (profit, people and planet) approach and consider business operations as well as environmental impacts and social responsibilities of the suppliers. Different metrics are introduced to measure performance in these three areas. To examine the influences of different organizational and supply chain operating philosophies, the objectives in selection of suppliers are designed so that some of them favor profit or the business operations, others the planet or the environment and the remaining focusing on people or social responsibility. A novel methodological approach based on a Bayesian framework and Monte Carlo Markov Chain (MCMC) simulation is developed to rank and select suppliers using specific selection objectives. This technique is also effective when smaller or missing data sets exist, which is an especially prevalent characteristic for newer and complex measures such as in a sustainability decision environment. Results obtained from the MCMC simulation provide a wealth of information about supplier performance, which form the basis for additional statistical analyses. The model allows the decision maker to execute various scenarios by changing importance weights attached to the triple-bottom-line areas. We present results for some of those scenarios with managerial and research implications and future research directions identified.

Scalable inference for Markov processes with intractable likelihoods

Bayesian inference for Markov processes has become increasingly relevant in recent years. Problems of this type often have intractable likelihoods and prior knowledge about model rate parameters is often poor. Markov Chain Monte Carlo (MCMC) techniques can lead to exact inference in such models but in practice can suffer performance issues including long burn-in periods and poor mixing. On the other hand approximate Bayesian computation techniques can allow rapid exploration of a large parameter space but yield only approximate posterior distributions. Here we consider the combined use of approximate Bayesian computation and MCMC techniques for improved computational efficiency while retaining exact inference on parallel hardware.

Cross-border sexual transmission of the newly emerging HIV-1 clade CRF51_01B.

A novel HIV-1 recombinant clade (CRF51_01B) was recently identified among men who have sex with men (MSM) in Singapore. As cases of sexually transmitted HIV-1 infection increase concurrently in two socioeconomically intimate countries such as Malaysia and Singapore, cross transmission of HIV-1 between said countries is highly probable. In order to investigate the timeline for the emergence of HIV-1 CRF51_01B in Singapore and its possible introduction into Malaysia, 595 HIV-positive subjects recruited in Kuala Lumpur from 2008 to 2012 were screened. Phylogenetic relationship of 485 amplified polymerase gene sequences was determined through neighbour-joining method. Next, near-full length sequences were amplified for genomic sequences inferred to be CRF51_01B and subjected to further analysis implemented through Bayesian Markov chain Monte Carlo (MCMC) sampling and maximum likelihood methods. Based on the near full length genomes, two isolates formed a phylogenetic cluster with CRF51_01B sequences of Singapore origin, sharing identical recombination structure. Spatial and temporal information from Bayesian MCMC coalescent and maximum likelihood analysis of the protease, gp120 and gp41 genes suggest that Singapore is probably the country of origin of CRF51_01B (as early as in the mid-1990s) and featured a Malaysian who acquired the infection through heterosexual contact as host for its ancestral lineages. CRF51_01B then spread rapidly among the MSM in Singapore and Malaysia. Although the importation of CRF51_01B from Singapore to Malaysia is supported by coalescence analysis, the narrow timeframe of the transmission event indicates a closely linked epidemic. Discrepancies in the estimated divergence times suggest that CRF51_01B may have arisen through multiple recombination events from more than one parental lineage. We report the cross transmission of a novel CRF51_01B lineage between countries that involved different sexual risk groups. Understanding the cross-border transmission of HIV-1 involving sexual networks is crucial for effective intervention strategies in the region.

Phylodynamic analysis reveals CRF01_AE dissemination between Japan and neighboring Asian countries and the role of intravenous drug use in transmission.

BackgroundOne major circulating HIV-1 subtype in Southeast Asian countries is CRF01_AE, but little is known about its epidemiology in Japan. We conducted a molecular phylodynamic study of patients newly diagnosed with CRF01_AE from 2003 to 2010.MethodsPlasma samples from patients registered in Japanese Drug Resistance HIV-1 Surveillance Network were analyzed for protease-reverse transcriptase sequences; all sequences undergo subtyping and phylogenetic analysis using distance-matrix-based, maximum likelihood and Bayesian coalescent Markov Chain Monte Carlo (MCMC) phylogenetic inferences. Transmission clusters were identified using interior branch test and depth-first searches for sub-tree partitions. Times of most recent common ancestor (tMRCAs) of significant clusters were estimated using Bayesian MCMC analysis.ResultsAmong 3618 patient registered in our network, 243 were infected with CRF01_AE. The majority of individuals with CRF01_AE were Japanese, predominantly male, and reported heterosexual contact as their risk factor. We found 5 large clusters with ≥5 members and 25 small clusters consisting of pairs of individuals with highly related CRF01_AE strains. The earliest cluster showed a tMRCA of 1996, and consisted of individuals with their known risk as heterosexual contacts. The other four large clusters showed later tMRCAs between 2000 and 2002 with members including intravenous drug users (IVDU) and non-Japanese, but not men who have sex with men (MSM). In contrast, small clusters included a high frequency of individuals reporting MSM risk factors. Phylogenetic analysis also showed that some individuals infected with HIV strains spread in East and South-eastern Asian countries.ConclusionsIntroduction of CRF01_AE viruses into Japan is estimated to have occurred in the 1990s. CFR01_AE spread via heterosexual behavior, then among persons connected with non-Japanese, IVDU, and MSM. Phylogenetic analysis demonstrated that some viral variants are largely restricted to Japan, while others have a broad geographic distribution.