Sean P. Murphy

Recombinant Temporal Aberration Detection Algorithms for Enhanced Biosurveillance

OBJECTIVEnBroadly, this research aims to improve the outbreak detection performance and, therefore, the cost effectiveness of automated syndromic surveillance systems by building novel, recombinant temporal aberration detection algorithms from components of previously developed detectors.nnnMETHODSnThis study decomposes existing temporal aberration detection algorithms into two sequential stages and investigates the individual impact of each stage on outbreak detection performance. The data forecasting stage (Stage 1) generates predictions of time series values a certain number of time steps in the future based on historical data. The anomaly measure stage (Stage 2) compares features of this prediction to corresponding features of the actual time series to compute a statistical anomaly measure. A Monte Carlo simulation procedure is then used to examine the recombinant algorithms ability to detect synthetic aberrations injected into authentic syndromic time series.nnnRESULTSnNew methods obtained with procedural components of published, sometimes widely used, algorithms were compared to the known methods using authentic datasets with plausible stochastic injected signals. Performance improvements were found for some of the recombinant methods, and these improvements were consistent over a range of data types, outbreak types, and outbreak sizes. For gradual outbreaks, the WEWD MovAvg7+WEWD Z-Score recombinant algorithm performed best; for sudden outbreaks, the HW+WEWD Z-Score performed best.nnnCONCLUSIONnThis decomposition was found not only to yield valuable insight into the effects of the aberration detection algorithms but also to produce novel combinations of data forecasters and anomaly measures with enhanced detection performance.

First experiences with semi-autonomous robotic harvesting of protein crystals

The demonstration unit of the Universal Micromanipulation Robot (UMR) capable of semi-autonomous protein crystal harvesting has been tested and evaluated by independent users. We report the status and capabilities of the present unit scheduled for deployment in a high-throughput protein crystallization center. We discuss operational aspects as well as novel features such as micro-crystal handling and drip-cryoprotection, and we extrapolate towards the design of a fully autonomous, integrated system capable of reliable crystal harvesting. The positive to enthusiastic feedback from the participants in an evaluation workshop indicates that genuine demand exists and the effort and resources to develop autonomous protein crystal harvesting robotics are justified.

The pandemic influenza policy model: a planning tool for military public health officials.

The Pandemic Influenza Policy Model (PIPM) is a collaborative computer modeling effort between the U.S. Department of Defense (DoD) and the Johns Hopkins University Applied Physics Laboratory. Many helpful computer simulations exist for examining the propagation of pandemic influenza in civilian populations. We believe the mission-oriented nature and structured social composition of military installations may result in pandemic influenza intervention strategies that differ from those recommended for civilian populations. Intervention strategies may differ between military bases because of differences in mission, location, or composition of the population at risk. The PIPM is a web-accessible, user-configurable, installation-specific disease model allowing military planners to evaluate various intervention strategies. Innovations in the PIPM include expanding on the mathematics of prior stochastic models, using military-specific social network epidemiology, utilization of DoD personnel databases to more accurately characterize the population at risk, and the incorporation of possible interventions, e.g., pneumococcal vaccine, not examined in previous models.

Data classification for selection of temporal alerting methods for biosurveillance

This study presents and applies a methodology for selecting anomaly detection algorithms for biosurveillance time series data. The study employs both an authentic dataset and a simulated dataset which are freely available for replication of the results presented and for extended analysis. Using this approach, a public health monitor may choose algorithms that will be suited to the scale and behavior of the data of interest based on the calculation of simple discriminants from a limited sample. The tabular classification of typical time series behaviors using these discriminants is achieved using the ROC approach of detection theory, with realistic, stochastic, simulated signals injected into the data. The study catalogues the detection performance of 6 algorithms across data types and shows that for practical alert rates, sensitivity gains of 20% and higher may be achieved by appropriate algorithm selection.

Minimalistic Predictor of Protein Binding Energy: Contribution of Solvation Factor to Protein Binding

It has long been known that solvation plays an important role in protein-protein interactions. Here, we use a minimalistic solvation-based model for predicting protein binding energy to estimate quantitatively the contribution of the solvation factor in protein binding. The factor is described by a simple linear combination of buried surface areas according to amino-acid types. Even without structural optimization, our minimalistic model demonstrates a predictive power comparable to more complex methods, making the proposed approach the basis for high throughput applications. Application of the model to a proteomic database shows that receptor-substrate complexes involved in signaling have lower affinities than enzyme-inhibitor and antibody-antigen complexes, and they differ by chemical compositions on interfaces. Also, we found that protein complexes with components that come from the same genes generally have lower affinities than complexes formed by proteins from different genes, but in this case the difference originates from different interface areas. The model was implemented in the software PYTHON, and the source code can be found on the Shakhnovich group webpage: http://faculty.chemistry.harvard.edu/shakhnovich/software.

A method for rapid simulation of propagating wave fronts in three-dimensional cardiac muscle with spatially-varying fiber orientations

We describe a simple cellular automata model for cardiac electrical activation that can simulate propagation in 3D muscle with arbitrary local fiber orientations. The model is computationally efficient and can provide activation sequences for rapidly driving models of cardiac mechanics. Here we show how to compute the model parameters from tissue parameters and successfully demonstrate rapid simulation of wave fronts in a slab of tissue with a rotating fiber structure.

An interdisciplinary approach to integrated modeling of human systems for spaceflight

Upon undertaking the visionary task of constructing a digital human for space physiology, it is prudent to focus on specific questions that might be answered by integrated models. For example, to address cardiovascular alterations risks during space travel, one can benefit from modeling the steady-state and dynamical responses of cardiovascular system variables (heart rate, blood pressure, etc.) following hypothetical perturbations of the systems parameters, such as cardiac electrical state changes. At a multi-system level, one may want to study effects of exercise by incorporating models of skeletal muscle, metabolism, etc. This paper reports on initial efforts to link multiple human system models using an interdisciplinary approach. In this effort, we have connected a medium-fidelity physics-based model of the left and right ventricles of the heart with an existing cardiovascular system simulation (RCVSIM), using the High Level Architecture (HLA) standard for simulation interoperability developed in the Department of Defense (DoD), to produce a Cardiovascular-Ventricular System (CVVS) federated simulation. The combined simulation allows hemodynamic consequences of an arrhythmia to be simulated with modest computational resources in a reasonable time. The paper also reports on extensions to make the simulation suitable for simulating and understanding cardiovascular responses to different exercise regimens.

Decoupling temporal aberration detection algorithms for enhanced biosurveillance

This study decomposes existing temporal aberration detection algorithms into two, sequential stages and investigates the individual impact of each stage on outbreak detection performance. The data forecasting stage (stage 1) generates a prediction of the value of the time series a certain number of time steps in the future based on historical data. The anomaly measure stage (stage 2) compares one or more features of this prediction to the actual time series to compute a measure of the potential anomaly. This decomposition was found not only to yield valuable insight into the effects of the aberration detection algorithms but also to produce novel combinations of data forecasters and anomaly measures with enhanced detection performance.