Lars Seemann

Social network analysis and agent-based modeling in social epidemiology

The past five years have seen a growth in the interest in systems approaches in epidemiologic research. These approaches may be particularly appropriate for social epidemiology. Social network analysis and agent-based models (ABMs) are two approaches that have been used in the epidemiologic literature. Social network analysis involves the characterization of social networks to yield inference about how network structures may influence risk exposures among those in the network. ABMs can promote population-level inference from explicitly programmed, micro-level rules in simulated populations over time and space. In this paper, we discuss the implementation of these models in social epidemiologic research, highlighting the strengths and weaknesses of each approach. Network analysis may be ideal for understanding social contagion, as well as the influences of social interaction on population health. However, network analysis requires network data, which may sacrifice generalizability, and causal inference from current network analytic methods is limited. ABMs are uniquely suited for the assessment of health determinants at multiple levels of influence that may couple with social interaction to produce population health. ABMs allow for the exploration of feedback and reciprocity between exposures and outcomes in the etiology of complex diseases. They may also provide the opportunity for counterfactual simulation. However, appropriate implementation of ABMs requires a balance between mechanistic rigor and model parsimony, and the precision of output from complex models is limited. Social network and agent-based approaches are promising in social epidemiology, but continued development of each approach is needed.

A complex systems approach to evaluate HIV prevention in metropolitan areas: preliminary implications for combination intervention strategies.

Background HIV transmission among injecting and non-injecting drug users (IDU, NIDU) is a significant public health problem. Continuing propagation in endemic settings and emerging regional outbreaks have indicated the need for comprehensive and coordinated HIV prevention. We describe the development of a conceptual framework and calibration of an agent-based model (ABM) to examine how combinations of interventions may reduce and potentially eliminate HIV transmission among drug-using populations. Methodology/Principal Findings A multidisciplinary team of researchers from epidemiology, sociology, geography, and mathematics developed a conceptual framework based on prior ethnographic and epidemiologic research. An ABM was constructed and calibrated through an iterative design and verification process. In the model, “agents” represent IDU, NIDU, and non-drug users who interact with each other and within risk networks, engaging in sexual and, for IDUs, injection-related risk behavior over time. Agents also interact with simulated HIV prevention interventions (e.g., syringe exchange programs, substance abuse treatment, HIV testing) and initiate antiretroviral treatment (ART) in a stochastic manner. The model was constructed to represent the New York metropolitan statistical area (MSA) population, and calibrated by comparing output trajectories for various outcomes (e.g., IDU/NIDU prevalence, HIV prevalence and incidence) against previously validated MSA-level data. The model closely approximated HIV trajectories in IDU and NIDU observed in New York City between 1992 and 2002, including a linear decrease in HIV prevalence among IDUs. Exploratory results are consistent with empirical studies demonstrating that the effectiveness of a combination of interventions, including syringe exchange expansion and ART provision, dramatically reduced HIV prevalence among IDUs during this time period. Conclusions/Significance Complex systems models of adaptive HIV transmission dynamics can be used to identify potential collective benefits of hypothetical combination prevention interventions. Future work will seek to inform novel strategies that may lead to more effective and equitable HIV prevention strategies for drug-using populations.

A Robust Topology-Based Algorithm for Gene Expression Profiling

Early and accurate diagnoses of cancer can significantly improve the design of personalized therapy and enhance the success of therapeutic interventions. Histopathological approaches, which rely on microscopic examinations of malignant tissue, are not conducive to timely diagnoses. High throughput genomics offers a possible new classification of cancer subtypes. Unfortunately, most clustering algorithms have not been proven sufficiently robust. We propose a novel approach that relies on the use of statistical invariants and persistent homology, one of the most exciting recent developments in topology. It identifies a sufficient but compact set of genes for the analysis as well as a core group of tightly correlated patient samples for each subtype. Partitioning occurs hierarchically and allows for the identification of genetically similar subtypes. We analyzed the gene expression profiles of 202 tumors of the brain cancer glioblastoma multiforme (GBM) given at the Cancer Genome Atlas (TCGA) site. We identify core patient groups associated with the classical, mesenchymal, and proneural subtypes of GBM. In our analysis, the neural subtype consists of several small groups rather than a single component. A subtype prediction model is introduced which partitions tumors in a manner consistent with clustering algorithms but requires the genetic signature of only 59 genes.

Using Effective Subnetworks to Predict Selected Properties of Gene Networks

Background Difficulties associated with implementing gene therapy are caused by the complexity of the underlying regulatory networks. The forms of interactions between the hundreds of genes, proteins, and metabolites in these networks are not known very accurately. An alternative approach is to limit consideration to genes on the network. Steady state measurements of these influence networks can be obtained from DNA microarray experiments. However, since they contain a large number of nodes, the computation of influence networks requires a prohibitively large set of microarray experiments. Furthermore, error estimates of the network make verifiable predictions impossible. Methodology/Principal Findings Here, we propose an alternative approach. Rather than attempting to derive an accurate model of the network, we ask what questions can be addressed using lower dimensional, highly simplified models. More importantly, is it possible to use such robust features in applications? We first identify a small group of genes that can be used to affect changes in other nodes of the network. The reduced effective empirical subnetwork (EES) can be computed using steady state measurements on a small number of genetically perturbed systems. We show that the EES can be used to make predictions on expression profiles of other mutants, and to compute how to implement pre-specified changes in the steady state of the underlying biological process. These assertions are verified in a synthetic influence network. We also use previously published experimental data to compute the EES associated with an oxygen deprivation network of E.coli, and use it to predict gene expression levels on a double mutant. The predictions are significantly different from the experimental results for less than of genes. Conclusions/Significance The constraints imposed by gene expression levels of mutants can be used to address a selected set of questions about a gene network.

Effective Models of Periodically Driven Networks

Circadian rhythms are governed by a highly coupled, complex network of genes. Due to feedback within the network, any modification of the systems state requires coherent changes in several nodes. A model of the underlying network is necessary to compute these modifications. We use an effective modeling approach for this task. Rather than inferred biochemical interactions, our method utilizes microarray data from a group of mutants for its construction. With simulated data, we develop an effective model for a circadian network in a peripheral tissue, subject to driving by the suprachiasmatic nucleus, the mammalian pacemaker. The effective network can predict time-dependent gene expression levels in other mutants.