2004 Breakout Sessions
A1 - The Automated Clinical Laboratory as a Team Player in the Automated Hospital
A2 – CaBIG Tissue Banks and Pathology Workspace
B1 – Integrative Cancer Research in caBIG: Data Analysis & Statistical Methods
B2 – Tissue Banks and Pathology Tools and Clinical Trials in caBIG: Honest Broker Services: Optimizing the De-Identification Process
C1 – Shared Pathology Informatics Network
C2 – Clinical Trials in caBIG: From Consensus to Practice: A Discussion of the Development, Implementation Diffusion of Standards-based Information Technology
C3 – Integrative Cancer Research in caBIG: Networks and Pathways
C4 – Integrative Cancer Research in caBIG: Bioinformatics of High-Throughput Proteomics Session
D1 – Dueling Hackers
D2 – Vocabularies and Common Data Elements in caBIG: Terminology Services as an Infrastructure for Shared Vocabularies
E1 – Laboratory Information Management Systems for Specialty Pathology Laboratories: Challenges and Opportunities.
E2 –Architecture in caBIG: Developing a Universal Grid to Support Cancer Research
A1 — The Automated Clinical Laboratory as a Team Player In the Automated Hospital
Robin A. Felder, PhD
This discussion will identify the types of automation available on the market and the issues laboratories should consider when initiating an automation project. Significant cost savings as well as increases in laboratory efficiency and management system. Many pre-analyzed in a timely manner. Patient identification, specimen labeling, sample tracking, and analysis are key steps in assuring error free clinical diagnostics. New technologies are also being developed to improve throughout the hospital. Once specimens arrive in the laboratory, modular automation systems are available that will provide a flexible approach to optimizing specimen throughput, analytical turnaround time, and return on investment.
As hospitals move toward an electronic medical record, laboratories can install intelligent test reporting systems, which includes auto verification and automatic test expansion (automatic ordering of repeat, reflex or add-on tests). There are significant benefits to using auto verification including a reduction in fatique associated with reviewing data, a reduction in errors from overlooked abnormal results, an increase in patient safety, and improved reporting to the computing systems to in crease their productivity and improve patient outcomes. An efficient automated laboratory will work in concert with other automated systems in the hospital to improve patient safety, and reduced hospital stays. In the future, the use of predictive testing in the home will delay or reduce the amount of medical intervention necessary for our aging populations.
A2 — caBIG Tissue Banks and Pathology Workspace
Ashokkumar Patel, MD and Michael Feldman, MD, PhD
This presentation will describe the Tissue Banking and Pathology Tools workspace within the Cancer Biomedical Informatics Grid (caBIG) project. The NCI funded initiative “caBIG” is designed to create an integrated biomedical informatics grid among 60 NCI designated cancer centers and research institutes. This integrated grid will provide the architecture, language and systems integration to allow centers to share data, resources and applications (http://cabig.nci.nih.gov/caBIG/). The project is designed around “Workspaces”, one of which is Tissue Banks and Pathology Tools. The other Workspaces include Clinical Trials Management, Integrative Cancer Research, Vocabularies and Common Data Elements and finally Architecture. Within each workspace are developer and adopter sites. In the Tissue Bank and Pathology Tools workspace, there are two developer sites (University of Pittsburgh and Washington University) and six adopter sites (University of Pennsylvania, University of Arizona, Dartmouth University, Thomas Jefferson University, and Northwestern University). This presentation will discuss and describe the goals and deliverables of the Tissue Banking and Pathology Tools Workspace efforts to develop a common database for tissue banking, as well as a set of mapping tools for centers with pre-existing tissue bank software. In addition, the manner in which a set of common data elements within the pathology workspace will facilitate information sharing between centers with tissue banks as well as between centers sharing paraffin archives will also be described. We will present and discuss integration of tools for UMLS encoding and de-identification within pre-existing AP-LIS systems from the Shared Pathology Informatics Network (SPIN, see http://spin.nci.nih.gov) group into this caBIG workspace. Finally, we will discuss how these efforts extend other projects seeking to integrate pathology resources between centers. One such partnership involves the Pennsylvania Cancer Alliance Bioinformatics Consortium (www.pcabc.upmc.edu) and two NCI organ specific tissue resources: the Cooperative Prostate Cancer Tissue Resource (www.prostatetissues.org) and the Cooperative Breast Cancer Tissue Resource (www.cbctr.ims.nci.gov).
B1 — Data Analysis and Statistical Methods
David B. Allison, PhD, Steve Marron, PhD and Robert Clarke, PhD, DSc
View Synchronized Slides - Marron
View Synchronized Slides - Clarke
An overview of data analysis and statistical methods in the era of high-throughput genomics and proteomics, with an emphasis on microarray data analysis, including:
- Challenges in normalization from the multivariate perspective
- Statistical methods for accounting for tissue heterogeneity in cancer samples
B2 — Tissue Banks and Pathology Tools and Clinical Trials in caBIG:
Honest Broker Services: Optimizing Research Information Services Through Creative
Expansion of the Cancer Registry
Sharon Winters, MS, RHIA, CTR and Susan Urda, CTR
View Synchronized Slides
Department of HHS and OHRP Guidance Document
Collaborative Honest Broker Service Application
The key objective of this session will be to reveal methods for developing a Collaborative Honest Broker Service in the Informatics environment to support Tissue Banking, Research Information Services and Clinical Trials Research. These collaborative efforts have resulted in the reduction of duplication of activities and streamlined the process for de-identification and integration of pathology, tissue bank and cancer registry data at the Hillman Cancer Center of the University of Pittsburgh Cancer Institute (UPCI) and University of Pittsburgh Medical Center (UPMC). An in-depth overview of the development and management of this collaborative service, examples of processes used in de-identification and tools used to monitor the use of data from multiple data sources will be presented.
Return to top of page
C1 — Shared Pathology Informatics Network
Rebecca S. Crowley, MD, MSIS, David A. Berkowicz, MD, and Kevin Mitchell, MS
View Synchronized Slides - Crowley
View Synchronized Slides - Mitchell
The Shared Pathology Informatics Network (SPIN) is an NCI-funded consortium project involving Harvard University, University of Indiana, University of Pittsburgh, and the University of California, Los Angeles. The purpose of the project is to develop methods and processes for searching and accessing surgical pathology reports and archived tissue across institutional boundaries. In this breakout session, we will discuss and demonstrate the SPIN peer-to-peer architecture, Java-based resources for converting free-text reports into concept-coded XML documents, and tools for querying the network. Future plans for the project will be outlined. The tools developed by SPIN are open-source, and are freely available for use by the community. Institutions are invited to use these tools, and to consider interacting with networks in the future.
C2 — Clinical Trials in caBIG:
From Consensus to Practice: A Discussion of the Development, Implementation, and Diffusion of Standards-based Information Technology
Joyce C. Niland, PhD and Douglas B. Fridsma, MD, PhD
View Synchronized Slides - Niland
View Synchronized Slides - Fridsma
As information technology matures, it becomes increasingly important that researchers not only create innovative information technology to capture health information, but also capture that information in a sharable, standards-based format. The real risk comes after consensus has been reached — how do you take the standards as they have been created and deploy them into real-world applications? And how do we develop and deploy metadata repositories that follow international standards and facilitate mapping of data and integration among systems?
In this panel discussion, we will review the development of national and international clinical research information models, describe the real-world experiences in developing and using common data elements for cancer research networks, and discuss some of the problems and barriers to the diffusion and adoption of standards-based information technology. Discussion will focus on what we can learn from previous work in this area, and steps that can be taken to move standards from consensus to use in the healthcare organizations.
C3 — Integrative Cancer Research in caBIG:
Networks and Pathways
Dynamic Construction of Pathway Networks
Carl Schaefer, PhD
The prototype Pathway Interaction Database (National Cancer Institute, Center for Bioinformatics, http://cmap.nci.nih.gov/PW) allows the dynamic construction and visualization of pathway networks from a database of individual interactions. Networks can be specified by a combination of means: identifying pre-defined pathways of interest, identifying molecules of interest, and extending the initially selected set of interactions by including upstream and downstream interactions including upstream and downstream interactions.
Inference of Genetic Networks from Mutant Phenotypes
Blaz Zupan, PhD
In the talk, we will show how genetic networks can be reconstructed from data on mutant phenotypes. The method employs em\xpert-defined patterns to uncover gene relations from the data, and uses these relations as constraints in the search for plausible genetic networks. We originally crafted the method for inference from morphological and other specific phenotypes, and are currently extending it to deal with gene expression phgenotypes. Both approaches, together with implementations within a Web-based system called GenePath (www.genepath.org) and a data mining suite called Orange (www.ailab.si/orange), will be presented.
Identifying Pathway Activity from High-Throughput Data
Michael Ochs, PhD
The cellular signaling pathways that control cell growth, differentiation, apoptosis, and motility play a critical role in many cancers. Microarray and proteomics technologies provide detailed information on mRNA and protein levels in vivo. However, because of the limited nature of our knowledge of signaling pathways in humans and high noise levels in the data, analysis is difficult. Here we present a Bayesian method to encode known biological knowledge in order to enhance the usefulness of expression and proteomics data for determination of signaling activity
C4— Bioinformatics of High-Throughput Proteomics
Michael Ochs, PhD, James Lyons-Weiler, PhD, Milos Hauskrecht, PhD, Simon Lin, MD, Harry Burke, MD, PhD, Jeffrey Morris, PhD, Zhen Zhang,PhD and Wei Zhu, PhD
The focus of this session will be to explore ways in which informaticians at distant sites can benefit by working together on a common bioinformatics analysis solution aimed at optimizing biomarker discovery in proteomic peptide time-of-flight (SELDI and MALDI-TOF) profiles. Panelists will include members of the caBIG initiative and the Early Research Detection Network. We will explore:
- Challenges in Accounting for Mass Drift in Proteomic Profiles
- Normalization and Baseline Subtraction
- Data Representation: What Should be Analyzed?
- Feature Selection Methods
- Construction and Validation of Classifiers
- Development standards for caProteo (a proposed tool for analysis of proteomic)
D1 — Dueling Hackers
James H. Harrison, Jr., MD, PhD and Paul G. Catrou, MD
“Hacking,” the art of rapidly developing small, useful computer programs can be an effective way to extend or add new features to clinical laboratory systems. These small programs can substantially contribute to laboratory workflow, reduce clerical time or enable new quality assurance activities by rapidly processing data extracted from clinical systems. Modern commercial and open source (free) programming environments offer relatively low learning curves, short software development times and quick debugging. Drs. Harrison and Catrou will simultaneously present and critique, in “point-counterpoint” fashion, their own ideas about software development in the laboratory including a survey of languages (C, C++, Java, M/MUMPS, Visual Basic, Python and Perl) and examples from their own experience. The session will finish with a demonstration of the development of a short program in Visual Basic and Python. By the close of the session, attendees will be able to:
• Understand important differences in available computer languages
• Prioritize the strengths and weaknesses of programming environments
• Understand how to begin to learn programming and the commitment required
• Find and choose online and print tutorial and reference resources
• Identify laboratory problems appropriate for small programming projects.
D2 — Vocabularies and Common Data Elements in caBIG:
Terminology Services as an Infrastructure for Shared Vocabularies
Christopher G. Chute, MD and Harold R. Solbrig
The problem of managing comparable and consistent information occurs across the spectrum of cancer center applications from basic science biology to clinical characterizations. Furthermore, the semantic linkage of concepts expressed within genomic findings to clinical manifestations poses challenging onto logic problems. We will overview traditional mechanisms for addressing these terminology and semantic linkage issues and represents how they might become manifest within the caBIG vocabulary workspace.
E1 — Laboratory Information Management Systems for Specialty Pathology Laboratories: Challenges and Opportunities.
Federico A. Monzon, MD and Michael Sendek
View Synchronized Slides - Monzon
View Synchronized Slides - Sendek
View Synchronized Slides - Carter
Specialty laboratories within Pathology departments have data management needs that are not usually supported by standard (out of the box) commercial LIS offerings. These needs often demand new solutions, as well as a combination of tools and resources from existing Anatomic Pathology and Clinical Pathology systems. This session will provide an overview of the information management needs for specialty laboratories like HLA, and Molecular Diagnostics and the not so distant integration of genomic and proteomic data into clinical systems. We will discuss the challenges posed by these needs, the current solutions with commercial and in-house developed applications and the development of "hybrid" LIMS systems for managing information from high-throughput laboratory technologies.
E2 — Architecture in caBIG:
Developing a Universal Grid to Support Cancer Research
Joel Saltz, MD, PhD
There are tremendous benefits that can be gained through coordinated multi-site clinical and basic research endeavors. The National Cancer Institute has tasked the cancer research community to develop architecture able to support coordinated studies that involve clinical, molecular and image data. We will provide an overview of emerging caBIG architectural components. This will include a description of grid service interfaces, how to expose data and metadata structures, how to advertise services, language and runtime support for execution of grid queries and workflow. We will also cover issues related to grid security, authentication, authorization and issues related to honest broker requirements.
Participants in the workshop will:
- Learn about the impact of data and computational grids on cancer research
- Understand technical basis underlying grid architectures
- Develop a working knowledge of how grid technology can be used to integrate cancer database resources located at multiple sites
- Obtain a working knowledge of issues related to grid security, authentication, authorization and honest broker requirements.