Engineering topics

Pullum, Laura L

Publications and source records attributed to Pullum, Laura L.

10 records

DOE OSTI2016

Integrating Symbolic and Statistical Methods for Testing Intelligent Systems Applications to Machine Learning and Computer Vision

Embedded intelligent systems ranging from tiny im- plantable biomedical devices to large swarms of autonomous un- manned aerial systems are becoming pervasive in our daily lives. While we depend on the flawless functioning of such intelligent systems, and often take their behavioral correctness and safety for granted, it is notoriously difficult to generate test cases that expose subtle errors in the implementations of machine learning algorithms. Hence, the validation of intelligent systems is usually achieved by studying their behavior on representative data sets, using methods such as cross-validation and bootstrapping.In this paper, we present a new testing methodology for studying the correctness of intelligent systems. Our approach uses symbolic decision procedures coupled with statistical hypothesis testing to. We also use our algorithm to analyze the robustness of a human detection algorithm built using the OpenCV open-source computer vision library. We show that the human detection implementation can fail to detect humans in perturbed video frames even when the perturbations are so small that the corresponding frames look identical to the naked eye.

Jha, Sumit Kumar↗

DOE OSTI2015

ORBiT: Oak Ridge Bio-surveillance Toolkit for Public Health Dynamics

With novel emerging infectious diseases being reported across different parts of the world, there is a need to build effective bio-surveillance systems that can track, monitor and report such events in a timely manner. Apart from monitoring for emerging disease outbreaks, it is also important to identify susceptible geographic regions and populations where these diseases may have a significant impact. The digitization of health related information through electronic health records (EHR) and electronic healthcare claim reimbursements (eHCR) and the continued growth of self-reported health information through social media provides both tremendous opportunities and challenges in developing novel public health surveillance tools. In this paper, we present an overview of Oak Ridge Bio-surveillance Toolkit (ORBiT), which we have developed specifically to address data analytic challenges in the realm of public health surveillance. In particular, ORBiT provides an extensible environment to pull together diverse, large-scale datasets and analyze them to identify spatial and temporal patterns for various bio-surveillance related tasks. We demonstrate the utility of ORBiT in automatically extracting a small number of spatial and temporal patterns during the 2009-2010 pandemic H1N1 flu season using eHCR data. These patterns provide quantitative insights into the dynamics of how the pandemic flu spread across different parts of the country. We discovered that the eHCR data exhibits multi-scale patterns from which we could identify a small number of states in the United States (US) that act as bridge regions contributing to one or more specific influenza spread patterns. Similar to previous studies, the patterns show that the south-eastern regions of the US were widely affected by the H1N1 flu pandemic. Several of these south-eastern states act as bridge regions, which connect the north-east and central US in terms of flu occurrences. These quantitative insights show how the eHCR data combined with novel analytical techniques can provide important information to decision makers when an epidemic spreads throughout the country. Taken together ORBiT provides a scalable and extensible platform for public health surveillance.

big data analytics↗

DOE OSTI2015

Sequential Pattern Mining of Electronic Healthcare Reimbursement Claims: Experiences and Challenges in Uncovering How Patients are Treated by Physicians

We examine the use of electronic healthcare reimbursement claims (EHRC) for analyzing healthcare delivery and practice patterns across the United States (US). We show that EHRCs are correlated with disease incidence estimates published by the Centers for Disease Control. Further, by analyzing over 1 billion EHRCs, we track patterns of clinical procedures administered to patients with autism spectrum disorder (ASD), heart disease (HD) and breast cancer (BC) using sequential pattern mining algorithms. Our analyses reveal that in contrast to treating HD and BC, clinical procedures for ASD diagnoses are highly varied leading up to and after the ASD diagnoses. The discovered clinical procedure sequences also reveal significant differences in the overall costs incurred across different parts of the US, indicating a lack of consensus amongst practitioners in treating ASD patients. We show that a data-driven approach to understand clinical trajectories using EHRC can provide quantitative insights into how to better manage and treat patients. Based on our experience, we also discuss emerging challenges in using EHRC datasets for gaining insights into the state of contemporary healthcare delivery and practice in the US.

Pullum, Laura L↗

DOE OSTI2015

Constructing Patient Specific Clinical Trajectories from Electronic Healthcare Reimbursement Claims using Sequential Pattern Mining

We examine the use of electronic healthcare reimbursement claims (EHRC) for analyzing healthcare delivery and practice patterns across the United States (US). By analyzing over 1 billion EHRCs, we track patterns of clinical procedures administered to patients with heart disease (HD) using sequential pattern mining algorithms. Our analyses reveal that the clinical procedures performed on HD patients are highly varied leading up to and after the primary diagnosis. The discovered clinical procedure sequences reveal significant differences in the overall costs incurred across different parts of the US, indicating significant heterogeneity in treating HD patients. We show that a data-driven approach to understand patient specific clinical trajectories constructed from EHRC can provide quantitative insights into how to better manage and treat patients.

Pullum, Laura L↗

DOE OSTI2014

Update on Common-Cause Failure Experience and Mitigation Practices

Experience in other industries has shown that digital technology can provide substantial benefits in terms of performance and reliability. However, the U.S. nuclear power industry has been slow to adopt the technology extensively in its instrumentation and control (I&C) applications because of inhibiting factors such as regulatory uncertainty, insufficient technological experience base, implementation complexity, limited availability of nuclear-qualified products and vendors, and inadequate definition of modernization cost recapture. Although there have been examples of digital technology usage in the nuclear power industry, challenges to the qualification of digital technology for high-integrity nuclear power plant (NPP) applications have severely constrained more widespread progress in achieving the benefits that are possible through the transition to digital. The U.S. Department of Energy (DOE) Office of Nuclear Energy (NE) established the Advanced Sensors and Instrumentation (ASI) technology area under the Nuclear Energy Enabling Technologies (NEET) Program to coordinate the instrumentation and controls (I&C) research across DOE NE and to identify and lead efforts to address common needs. As part of the NEET ASI research program, the Digital Technology Qualification project was established. Under this project, the Oak Ridge National Laboratory (ORNL) is leading the investigation into mitigation of digital common-cause failure (CCF) vulnerabilities for nuclear-qualified applications. This technical report documents updated and expanded findings from research activities by ORNL. Specifically, the report describes CCF experience in the nuclear and nonnuclear industries, identifies the state of the practice for CCF mitigation through key examples, and presents conclusions from the determination of knowledge gaps.

Wood, Richard Thomas↗

DOE OSTI2014

Disease Prediction Models and Operational Readiness

INTRODUCTION: The objective of this manuscript is to present a systematic review of biosurveillance models that operate on select agents and can forecast the occurrence of a disease event. One of the primary goals of this research was to characterize the viability of biosurveillance models to provide operationally relevant information for decision makers to identify areas for future research. Two critical characteristics differentiate this work from other infectious disease modeling reviews. First, we reviewed models that attempted to predict the disease event, not merely its transmission dynamics. Second, we considered models involving pathogens of concern as determined by the US National Select Agent Registry (as of June 2011). Methods: We searched dozens of commercial and government databases and harvested Google search results for eligible models utilizing terms and phrases provided by public health analysts relating to biosurveillance, remote sensing, risk assessments, spatial epidemiology, and ecological niche-modeling, The publication date of search results returned are bound by the dates of coverage of each database and the date in which the search was performed, however all searching was completed by December 31, 2010. This returned 13,767 webpages and 12,152 citations. After de-duplication and removal of extraneous material, a core collection of 6,503 items was established and these publications along with their abstracts are presented in a semantic wiki at http://BioCat.pnnl.gov. Next, PNNL’s IN-SPIRE visual analytics software was used to cross-correlate these publications with the definition for a biosurveillance model resulting in the selection of 54 documents that matched the criteria resulting Ten of these documents, However, dealt purely with disease spread models, inactivation of bacteria, or the modeling of human immune system responses to pathogens rather than predicting disease events. As a result, we systematically reviewed 44 papers and the results are presented in this analysis.

biosecurity↗

DOE OSTI2014

Discovery of Disease Co-occurrence Patterns from Electronic Healthcare Reimbursement Claims Data

Effective public health surveillance is important for national secu- rity. With novel emerging infectious diseases being reported across different parts of the world, there is a need to build effective bio- surveillance systems that can track, monitor and report such events in a timely manner. Additionally, there is a need to identify sus- ceptible geographic regions/populations where these diseases may have a significant impact and design preemptive strategies to tackle them. With the digitization of health related information through electronic health records (EHR) and electronic healthcare claim re- imbursements (eHCR), there is a tremendous opportunity to ex- ploit these datasets for public health surveillance. In this paper, we present our analysis on the use of eHCR data for bio-surveillance by studying the 2009-2010 H1N1 pandemic flu season. We present a novel approach to extract spatial and temporal patterns of flu in- cidence across the United States (US) from eHCRs and find that a small, but distinct set of break-out patterns govern the flu and asthma incidence rates across the entire country. Further, we ob- serve a distinct temporal lag in the onset of flu when compared to asthma across geographic regions in the US. The patterns extracted from the data collectively indicate how these break-out patterns are coupled, even though the flu represents an infectious disease whereas asthma represents a typical chronic condition. Taken to- gether, our approach demonstrates how mining eHCRs can provide novel insights in tackling public health concerns.

data analytics↗

DOE OSTI2013

Integrating Heterogeneous Healthcare Datasets and Visual Analytics for Disease Bio-surveillance and Dynamics

n this paper, we present an overview of the big data chal- lenges in disease bio-surveillance and then discuss the use of visual analytics for integrating data and turning it into knowl- edge. We will explore two integration scenarios: (1) combining text and multimedia sources to improve situational awareness and (2) enhancing disease spread model data with real-time bio-surveillance data. Together, the proposed integration methodologies can improve awareness about when, where and how emerging diseases can affect wide geographic regions.

Ramanathan, Arvind↗

DOE OSTI2013

Oak Ridge Bio-surveillance Toolkit (ORBiT): Integrating Big-Data Analytics with Visual Analysis for Public Health Dynamics

In this position paper, we describe the design and implementation of the Oak Ridge Bio-surveillance Toolkit (ORBiT): a collection of novel statistical and machine learning tools implemented for (1) integrating heterogeneous traditional (e.g. emergency room visits, prescription sales data, etc.) and non-traditional (social media such as Twitter and Instagram) data sources, (2) analyzing large-scale datasets and (3) presenting the results from the analytics as a visual interface for the end-user to interact and provide feedback. We present examples of how ORBiT can be used to summarize ex- tremely large-scale datasets effectively and how user interactions can translate into the data analytics process for bio-surveillance. We also present a strategy to estimate parameters relevant to dis- ease spread models from near real time data feeds and show how these estimates can be integrated with disease spread models for large-scale populations. We conclude with a perspective on how integrating data and visual analytics could lead to better forecasting and prediction of disease spread as well as improved awareness of disease susceptible regions.

Ramanathan, Arvind↗

DOE OSTI2013

Extreme Scale Visual Analytics

Given the scale and complexity of today s data, visual analytics is rapidly becoming a necessity rather than an option for comprehensive exploratory analysis. In this paper, we provide an overview of three applications of visual analytics for addressing the challenges of analyzing climate, text streams, and biosurveilance data. These systems feature varying levels of interaction and high performance computing technology integration to permit exploratory analysis of large and complex data of global significance.

Steed, Chad A↗