As a result of comments received during the 2014 Joint Summits Town Hall meeting, we created a new format for 2015 with a specific focus on "Resources for Clinical and Translational Research". The new Peer-based Learning Workshops are designed for demonstrations and discussions of operational information systems used in clinical and translational research. The focus of these workshops is to encourage participation by researchers and practitioners from national initiatives, such as the Clinical and Translational Science Awards (CTSA), Electronic Medical Records and Genomics (eMERGE), the national Patient Centered Clinical Research Network (PCORNet), Newborn Screening Translational Research Network (NBSTRN), UK Clinical Research Network (UKCRN), NCI National Clinical Trials Network (NCTN), Rare Disease Clinical Research network (RDCRN), etc., with an emphasis on tools and resources as topics and hands-on interaction with attendees.
Tuesday, March 24
8:30 a.m. – 10:00 a.m.
TBI07: Use of the OMOP Common Data Model within the PCORI CDRN Initiative: Data Harmonization and Integration with PCORNet CDM
M. Matheny, Vanderbilt University/TVHS Veterans Administration; D. Meeker, University of Southern California; P. Ryan, Janssen Research & Development; L. Frey, Medical University of South Carolina; R. Belenkaya, Albert Einstein College of Medicine; M. Kahn, University of Colorado
The PCORI Clinical Data Research Network initiative is in the process of integrating electronic health record and directly reported patient information from 11 participating networks of healthcare systems to pursue large, distributed observational cohort studies and pragmatic clinical trials. Each network has a unique starting point regarding design and implementation of their EHR and clinical data warehouse resources. A collaborative effort between OHDSI and 3 CDRNs (pScanner, PEDSnet, and NYC) with existing or in-progress deployments of the OMOP CDM have been developing crosswalks to the PCORNet CDM version 1 for OMOP version 4 and 5 in order to integrate with the emerging PCORNet CDM standard. This workshop seeks to 1) provide a general overview of established common data models (I2B2, OMOP, PCORNet, Quality Data Model, and others) including relative strengths and weaknesses for categories of applications, 2) experience and challenges of transforming EHR data to the OMOP common data model, 3) ongoing work to provide cross-CDM interfaces between OMOP and PCORNet, and 4) development of ETL tools to help accelerate the transformation process for healthcare systems starting or in the process of data transformation, and 5) discussion of the process and ongoing efforts to update and support expanding use of a CDM from the curator’s perspective (OHDSI). In summary, this workshop is intended for audiences interested in using or implementing a secondary data use common data model within their institution, and serves as both an introduction to these types of models, and then proceeding into more detail with regards to the use of the PCORNet and OMOP CDMs for use within the CDRN network. Lessons learned and best practices will be discussed, tools being developed will be highlighted to help systems and groups perform these tasks more easily, and key challenges in application and re-use of CDMs will be described.
10:30 a.m. – 12:00 p.m.
TBI11: The Monarch Initiative: Semantic Phenotyping for Translational Medicine
M. Haendel, Oregon Health & Sciences University; N. Washington, C. Mungall, Lawrence Berkeley National Laboratory
It is well-known that mutations in orthologous genes and genes in the same signaling pathway often manifest in similar phenotypes, and therefore study of variant phenotypes in model systems may provide insight into human gene function and understanding of disease. Traditionally, basic researchers and clinical geneticists have searched for and integrated this knowledge manually. However, with the rapid rise in output of genomic sequencing technologies and high-throughput phenotyping efforts, manual identification, integration, and evaluation of relevant phenotype data will quickly become intractable. Further, such data is difficult to interpret, requiring domain knowledge of each species’ genetics, development, and even the specialized vocabulary used to describe anatomy, phenotype, and genotype.
Furthermore, each organism is best suited for study of different biological phenomena, meaning an integrated view over all organisms is best for understanding relevance to human disease. Until recently, there has been a deficiency of tools that leverage cross-species phenotypes in translational research. The Monarch Initiative was created to provide an integrated data portal and suite of tools to address this need. This workshop will cover the use of tools that leverage semantic phenotyping for the exploration of genotype-phenotype correlations for exome analysis, such as PhenIX and Exomiser, to aid disease-gene discovery. We will also focus the development and use of ontologies for capturing variant phenotypes and the evaluation of their quality using a phenotypic profile “sufficiency meter.” Finally, we will highlight the visualization of human disease phenotypes in comparison to model systems to aid variant validation and model selection. We encourage participants to bring questions about the use of ontologies to capture phenotypes, VCF files together with phenotype annotations for analysis with Exomiser and PhenIX, and exploratory questions about their favorite animal model, disease, or gene.
Thursday, March 26
7:00 a.m. – 10:00 a.m.
CRI10: Visualizing Business Analytics and HIT for PCORI
Use of Innovative HIT Solutions in Clinical Data Research Networks for Optimization of Patient-centered Clinical Outcomes Research
M. Zirkle, PCORI; A.N. Kho, Northwestern University; R. Waitman, University of Kansas Medical Center; T. Carton, Louisiana Public Health Institute; P. Harris, Vanderbilt University Medical Center; L. Ohno-Machado, University of California, San Diego; K. Mandl, Harvard Medical School/Boston Children's Hospital
PCORnet, the National Patient-Centered Clinical Research Network, is an innovative initiative of the Patient-Centered Outcomes Research Institute (PCORI). The goal of PCORnet is to improve the nation’s capacity to conduct comparative effectiveness research efficiently by creating a large, highly representative network for conducting clinical outcomes research. By requiring that the health systems, clinicians, and patients who generate data be involved in all aspects of the governance and use of those data, PCORnet aims to facilitate the shift in health research from researcher-driven to patient-centered research. PCORnet brings together the expertise, populations, resources, and data of its participating organizations to create a national infrastructure that enables more efficient, patient-centered research. PCORnet includes 11 Clinical Data Research Networks (CDRNs), which are research networks based in healthcare delivery systems. Each CDRN contributes to the infrastructure for this national network by way of innovative informatics solutions to conduct patient-centered outcomes research. PCORnet is blending these informatics solutions to harness big data with the optimization of patient engagement in building our infrastructure, thus highlighting the opportunities for informatics to enable research. PCORnet is a highly collaborative community that incubates and shares novel solutions for IT-enabled research, with an emphasis on the distinctive tools and resources created by the CDRNs. We will create a peer-to-peer environment suitable for learning, engagement and ideal interaction among all of the attendees.
Visualizing Business Analytics in Biomedical Informatics
L. McIntosh, C. Zabarovskaya, Washington University School of Medicine
With the advancement of modern technology, measuring performance of systems and processes has become an important element of business operations. To improve efficiency and implement accountability within academic cores, we need ready access to information from operational data in order to assess (academic) business analytics. The answer to this is often a real-time dashboard or information monitoring system. However, before committing to a tool or a chart type, one must first identify a message for their visualization, preferably one that will be actionable. We will walk you through how to create one in R and why we think this is a great way to start. This workshop will address both theories and methods of building charts and dashboards in an open-source environment. Participants of this workshop will first learn the intricacies of story-telling using graphics and how to identify useful Key Performance Indicators (KPIs). Next, best practices in dashboard designs and an overview of the grammar of graphics will be covered.
We will move to hands-on programming opportunities using R software, rCharts package, and Shiny web app framework. These programs are being intensively contributed to and supported in the open-source community; they also provide a solid platform for customization. Finally, the workshop instructors will present various ways of publishing charts independently online or within a particular network (using Shiny Server). Participants will learn the basics of building a dashboard in Shiny and will be exposed to some of the advanced examples of analytical Shiny apps. Thus, by the end of the workshop, participants will be able to use open-source technologies for powerful graphics and actionable insights through: developing KPIs, building powerful visualizations using best-practices in information design, and sharing dashboards with various audiences.
3:30 p.m. – 5:00 p.m.
CRI26: Peer-based Learning Workshop - IRBShare and Honest Broker
IRBshare: Enabling Flexibility while Streamlining IRB Oversight and Administrative Management of Multicenter Trials
P. Harris, E. Sheffer, Vanderbilt University
Multisite research studies are frequently delayed due to well-known and publicized process-related inefficiencies in the IRB approvals process (Greene, Geiger, & Harris, 2006). Moreover, the value proposition for multiple, redundant site-specific IRB reviews is questionable. IRBshare (www.irbshare.org) is a joint IRB review model for multisite studies that facilitates the sharing of IRB-approved documents between IRBs and minimizes the need for all sites to conduct a full board review—thereby accelerating the IRB approval process. IRBshare is a consortium-based project with 45 participating institutional partners across the United States. The unique sharing and reliance model was developed with input from the Association for the Accreditation of Human Research Protection Programs (AAHRPP), the US Health and Human Services Office for Human Research Protections (OHRP), academic medical center regulatory and research experts and government/industry sponsors. During the consortium-building and platform pilot project phase (2 years of operation), IRBshare supported 11 clinical research studies and had a preliminary turnaround time of 18 days from submission to approval. We are currently expanding technical infrastructure support to accommodate multiple IRB review models – including shared review, central review, and local review – ultimately providing a real-world platform for capturing workflow process data in the course of study reviews and operations (enhanced platform: IRBchoice). We hypothesize that empowering institutions with the opportunity to choose the IRB model that contextually works best (based on institutional and study characteristics) will encourage greater utilization and ultimately improve the ethical quality, efficaciousness and efficiency of multisite study IRB reviews. For this panel, we will focus on broad, consortium-informed lessons learned related to improving the regulatory review process for multi-site studies, current and future technical platform considerations for IRBshare, and potential ideas for collaboration within the informatics community related to bi-directional project data transfer between local IRB systems and IRBshare.
Implementation of an Electronic Honest Broker-based Tool-kit for Integrated Data and Biorepository Research
A. Felmeister, Drexel University; T. Rivera, The Children's Hospital of Philadelphia
The Children’s Hospital of Philadelphia (CHOP) Center for Biomedical Informatics (CBMi) has developed an open-source informatics infrastructure to enable data-intensive longitudinal biorepository studies. Requirements met by the system include the need to de-identify and re-identify subjects on-the-fly and create and supplement records in multiple research systems to ensure patient privacy. Implementations of this infrastructure produce a longitudinally updated specimen and data resource that enables more flexible non-human subjects research. Our workshop will focus on the main challenges and concepts pervasive in a modern biorepository by utilizing our multi-system approach to management of these studies. The tools utilized and configured directly by participants in the workshop will include an electronic Honest Broker, a Research and Specimen Coordinator Portal that include a custom REDCap client. The final activity will be configuration and utilization of a Harvest* data discovery application (harvest.research.chop.edu) to expose the integrated data set created from this tool kit. Attendees will assume multiple roles including an informatics analyst, a specimen and data coordinator and finally a biomedical researcher while configuring each component of the integrated toolkit. They will collect patient identifiers to be stored in an electronic Honest Broker, enter specimen annotations (phenotypic data) utilizing a connected REDCap client, and associate specimen collection kits with the subject record. We will continue into a more comprehensive demonstration of integration with a the specimen-management application programming interface of a proprietary LIMS as that we will be unable to deploy to the attendees. Finally attendees will be able to interact with their annotated specimen data by configuring and deploying a harvest application as a data and specimen query tool. We will also demonstrate this tool on more complete data set than will be utilized for the workshop to show results. After the hands-on portion of the workshop, we will lead a discussion on the technologies used, focusing heavily on the utilization of the REDCap API for the Research Portal client. We will discuss the challenges and approaches of extract transform and load (ETL) tools and techniques used to bring information together. We will also dive into some of our lessons-learned on taking a modular approach including containerization of components using Docker to actually create the hands on workshop and do testing of software we develop.
Friday, March 27
8:30 a.m. – 10:00 a.m.
CRI31: Everything about Scientific Profiles: Resource Sharing and Networking
eagle-i : a VIVO-ISF Ontology enabled tool to promote resource sharing, discovery and integration with Harvard Catalyst Profiles
B. Bahl, D. Bourges-Waldegg, G. Weber, Harvard Medical School; M. Haendel, Oregon Health Science university; K. Holmes, Northwestern University
Developing research resources is time consuming and expensive, and finding resources for use in biomedical science is a challenge. A system that can promote sharing, reuse, and attribution for development of research resources such as cell lines and reagents could potentially increase efficiency and collaboration. eagle-i is a national network for sharing biomedical resources accross 35 institutions. In this workshop, we will first discuss how institutions can join the national network. As a case study, we will detail the usage of eagle-i software to power a statewide resource finding tool for North Carolina that also feeds resources into the national network. We will present a recent initiative to enrich the eagle-i network with a comprehensive catalog of induced pluripotent stem cell lines (iPSC). A partnership with the New York Stem Cell Institute (NYSCF) resulted in the inclusion of NYSCF’s vast collection of stem cell lines, and in the development of an advanced query interface. One of the goals of eagle-i is to support attribution for research resources. We will discuss the integration of the eagle-i ontology with the VIVO ontology to form the VIVO_ISF (Integrated Semantic Framework), which supports the representation of research resources as scholarly products. The VIVO-ISF ontology can be used as a research profile exchange standard, to support sharing and exchange of research expertise data, and aid attribution for roles played in the creation of scholarly products of any kind. Towards this end, we have integrated eagle-i content into the researcher networking tool Harvard Catalyst Profiles. With this integration, HC Profiles displays a short summary of resources that a researcher has shared in eagle-i.
Profiles Research Networking Software
G. Weber, Harvard Medical School; E. Meeks, University of California San Francisco
Profiles Research Networking Software (RNS) (http://profiles.catalyst.harvard.edu) is an open source tool originally developed by Harvard’s clinical and translational science center to help investigators find new collaborators. It automatically generates online profiles for an institution’s researchers by integrating data from local administrative systems with information from external sources, such as PubMed and NIH ExPorter. Using data mining algorithms, it discovers different types of networks that exist within an organization, such as people who have co-authored publications, who are conducting similar research, or whose offices are physically close. This information is displayed in numerous data visualizations to help users understand these networks. Profiles RNS is a Semantic Web application, which uses Linked Open Data (LOD) and the VIVO ontology to exchange data with other research networking (RN) systems and participate in federated expertise networks, such as Direct2Experts. UCSF added support for Open Research Networking Gadgets (ORNG) to Profiles RNS, which combines LOD and the OpenSocial standard (used by Google, IBM, LinkedIn, and others) to enable developers to create light-weight but powerful plug-ins for Profiles RNS. Profiles RNS is a free application under a BSD license, and optional commercial support services are available through the company Symplectic. This workshop will provide a broad overview of Profiles RNS, both discussing features common to many RN products as well as highlighting functionality unique to Profiles RNS. It will then drill down into two specific topics. The first describes how to use Google Analytics to evaluate an RN website, and then how use this information to improve search engine ranking and to encourage investigators to use the site. The second shows how bibliometric and social network analysis can be used to learn about an organization and design interventions that help investigators assemble more effective collaborations.