2017 Joint Summits Panels

Monday, March 27, 2017

10:30 a.m. – 12:00 p.m.

S01: Panel – Information Technology Powering Cancer Research for Discovery and Novel Hypothesis Generation
S. Madhavan, Georgetown University; J. Saltz, Stony Brook University; G. Jiang, Mayo Clinic; H. Xu, University of Texas Health Science Center at Houston

There is a critical need to develop research-driven informatics technology across the development lifecycle to address priority needs in cancer research and management. These technology platforms are required to support study design, data collection, processing, analysis, interpretation, clinical decision support and novel hypothesis generation. The National Cancer Institute is supporting various technology platforms in these contexts to drive cancer data sharing, clinical translational research, promote interdisciplinary science among other goals. This thematic panel will highlight four projects that are developing technology infrastructure and tools for cancer research in areas of clinical genomics, image analysis, natural language processing and data standards.

In this panel, we highlight four cancer technology development projects that describe various platforms that demonstrate the use of computational methods, tools and framework to drive discovery in the era of rapidly accumulating big data.

S02: Panel – Big Data for Pharmacovigilance: Challenge and Opportunity
P. Zhang, IBM T.J. Watson Research Center; G. Gonzalez, Arizona State University; R. Harpaz, Oracle Health Sciences; Y. Li, IBM T.J. Watson Research Center; N. Shah, Stanford University

Adverse drug reactions (ADRs) are a major burden for patients and the healthcare industry. It usually causes preventable hospitalizations and deaths, while associated with a huge amount of cost. Spontaneous reporting systems (SRSs) have been the cornerstone in pharmacovigilance for a long time, and are effective at detecting many types of ADRs. However, significant under-reporting bias inherently leaves patients at risk until sufficient clinical evidence has been granted. To augment the current systems, there are new ways to conduct pharmacovigilance using expanded data sources including data from electronic health records (EHRs), scientific literature and social media. Collectively labeled as the big data, they share the characteristics of large volumes, diversity and complexity that present both challenges and opportunities to its holders. Recently, the research community has devoted much effort to this field that may fundamentally transfer the manner in which ADR can be identified and evaluated. However, there is clearly room for technical improvement with regard to computational drug safety surveillance methods. Furthermore, to materialize the true potential and impact of these methods, much work is needed to show that they can be successfully adopted into practical applications. In this panel, participants will summarize the recent advances in big data for pharmacovigilance and identify challenges and opportunities. Panel participants will synthesize their perspectives on these key issues and vision the future developments in this area, explore a diverse set of topics, and engage in thoughtful discussion with the audience.

3:30 p.m. – 5:00 p.m.

S08: Panel – NIH Data Repositories and Cloud Resources: NCI Genomic Data Commons, NCI Cancer Genomics Cloud Pilots and NIH Commons Credits
T. Davidsen, National Cancer Institute; G. Komatsoulis, National Center for Biotechnology Information; J. Chuang, The Jackson Laboratory for Genomic Medicine, University of Connecticut Health Center; V. Thorsson, Institute for Systems Biology; J. Klemm, National Cancer Institute

The NIH has funded several resources in response to the growth of large-scale sequence data which is rapidly out-stripping the required computational capacity for storage, processing, network transmission, and analysis. First, the newly launched NCI Genomic Data Commons (GDC) allows users to access cancer genomics data from public projects, including The Cancer Genome Atlas (TCGA), and to submit their own data for community use. Second, the NCI Cancer Genomics Cloud (CGC) Pilots are three projects that each provide access to TCGA data with co-located computational capacity. They also each provide an Application Programming Interface (API) that supports direct data access and computation for developers of analytic tools. Third, the NIH Commons is offering the opportunity for NIH grantees to apply for cloud credits through the Commons Credits Model. These credits can be used by recipients to analyze their data on compliant commercial and academic cloud providers. During this panel we will introduce these three new resources and highlight the experience of two genomics researchers who have utilized and worked with the CGC Pilots and the GDC. The panel will conclude with a thirty-minute roundtable of all speakers addressing any questions and discussion topics from the audience.

Tuesday, March 28, 2017

10:30 a.m. – 12:00 p.m.

S11: Panel – Clinical Genomics and Big Data Research at the National Institute of Allergy and Infectious Diseases
S. Xirasagar, A. Yao, M. Quinones, D. Lin, National Institute of Allergy and Infectious Diseases

The mission of the National Institute of Allergy and Infectious diseases, NIAID, is to lead research to understand, treat, and prevent infectious, immunologic, and allergic diseases. Omics-based initiatives in basic, preclinical, and clinical research are rapidly replacing other traditional approaches to support the etiology and development of new diagnostics, prevent measures, vaccines, and therapeutics for infectious, immunologic, and allergic diseases. NIAID is committed to supporting the development of innovative, scalable, extensible, and feasible bioinformatics and computational biology informatics solutions to reap the full benefits of these Big Data and accelerate translation of research to improve human health. The four proposed panel experts will highlight intramural and extramural NIAID omics-based research, Big Data opportunities and challenges arising from such data, and informatics approaches and initiatives. Resources developed include those to address challenges in data management, analysis and sharing. Databases, tools to facilitate the use of molecular data in epidemiology, surveillance, metagenomics, population genetics, genotype/phenotype association, functional genomics, studies to understand the etiology of antimicrobial resistance and autoimmune disorders, host responses, host-pathogen interactions will be discussed.

S14: Panel – Mental Health Informatics
L. Rozenblit, Prometheus Research, LLC; P. Ranallo, Six Aims for Behavorial Health; D. Stark, Icahn School of Medicine at Mount Sinai; T. McCoy, Harvard University/Massachusetts General Hospital

Mental health disorders consist of more than 200 classified forms of mental illnesses, and account for more than $201 billion in expenditure in the U.S. (in 2013). They represent an extraordinary opportunity for informatics methodologies to have a positive impact, but also pose a unique set of challenges to be overcome. In this panel, speakers will describe both opportunities and challenges in research, practice, and policy as they relate to the mental health domain.

1:30 p.m. – 3:00 p.m.

S15: Panel – Making Precision Oncology Data More Usable for Research and Care 
S. Madhavan, Georgetown University; R. Ratwani, MedStar Institute for Innovation; J. Warner, Vanderbilt University; S. Bhavnani, University of Texas Medical Branch

Precision medicine in oncology implements clinical screening and tissue molecular profiling to characterize the genetic makeup of the patient (i.e., germline DNA) and of the tumor (i.e., somatic mutations). This consequently enables the identification and validation of treatments to reduce side effects, and improve outcomes. For example, current treatment for a breast cancer patient who has estrogen receptor positive (ER+) breast cancer will potentially include tamoxifen, a drug shown to be effective against early stage receptor-positive cancers. Therefore, the knowledge derived from both a patient’s phenotypic and molecular profile has the potential to support clinical decision-making by understanding the likely benefits and risks of a particular treatment.

For cancer patients progressing past first or second line ‘standard of care’ treatments, molecular profiling (MP) is often a last resort and the tumors considered are often from secondary, metastatic sites. The genes in these molecular profiling panels are specifically chosen because they encode known or likely targets of therapies, either approved or in clinical trials, or are known drivers of oncogenesis. Molecular profiling results report not only FDA-approved therapies intended for the patient’s tumor type, but also FDA-approved therapies intended for other tumor types (off-label applications) and clinical trials.

In this timely panel, we highlight challenges in consuming MP data for clinical decision making and novel hypothesis generation. The speakers will preset various projects that are standardizing evidence for associating MP data to cancer treatments, application of human factors to cancer molecular diagnostic reports and novel patient stratification approaches using data visualization techniques.

S18: Panel – Is an Algorithm Just an Algorithm? How to Define Clinical Concepts for Use in Comparative Safety and Effectiveness Distributed Research Networks
J. Brown, Harvard Pilgrim Health Care Institute; R. Ball, Food and Drug Administration; D. Carrell, Group Health Research Institute; L. D'Avolio, Harvard Medical School ; K. Marsolo, Children’s Hospital Research Foundation 
 
Real-world electronic health data are often used to generate evidence of real-world comparative safety and effectiveness of medical interventions. Coinciding with the widespread adoption of EHR systems is the proliferation of distributed research networks designed to combine data across institutions for secondary purposes, bringing into focus the potential for a learning health system. But this potential can only be realized if the research community can efficiently use the available electronic data to generate valid and robust evidence. A critical need for developing valid evidence relates to development of well-characterized electronic algorithms of clinical concepts (e.g., phenotypes, health outcomes/events of interest, or exposures of interest). Even with advances in algorithm development (e.g., via use of using natural language processing and machine learning), there is a growing need to improve methods for developing and validating algorithms, and to better understand how and when algorithms can be used appropriately. The panel will focus on the emerging needs, issues and opportunities for identifying and implementing algorithms in distributed networks. Attendees will learn about different approaches for developing and characterizing electronic algorithms, opportunities to speed algorithm development, and how different use cases effect algorithm selection and use.

Wednesday, March 29, 2017

8:30 a.m. – 10:00 a.m.

S22: Panel – A Perspective on Genomics in Clinical Practice: Focus on Pediatric Cancers
S. Volchenboum, University of Chicago; S. Meshinchi, Fred Hutchinson Cancer Center; R. Aplenc, Children's Hospital of Philadelphia ; E. Crowgey, Nemours Alfred I. duPont Hospital for Children; T. Druley, Washington University

This panel consists of leaders in the Children Oncology Group (COG) from several participating institutions. The panelists have a diverse range of expertise and will discuss the use of genomics for precision care and the need for a multidisciplinary team of clinicians and bioinformaticians. The moderator for the panel will be Samuel Volchenboum, the chair of AMIA’s Genomics and Translational Bioinformatics Working Group. The learning objectives from the panel will draw upon their experience with clinical trials and the application of genomic data and will include (1) identifying children with cancer who are most likely to benefit from genomic testing, (2) counseling patients and their caregivers prior to and after testing, and (3) interpreting and acting on results of genomic testing.

S25: Panel – Developing Governance Structures for Informatics Approaches to Research Recruitment
M. Cantor, NYU Langone Medical Center; P. Embi, Regenstrief Institute, Inc.; A. Hsiao, Yale School of Medicine

EHRs have great potential as a tool for recruiting patients into clinical studies. Many institutions are providing resources so that both investigators and patients can take full advantage of this potential, from identifying cohorts to sending patients alerts about trials in disease areas in which they may be interested. A well-defined governance process is essential for making these tools satisfy the needs of their end-users while also fitting into the current regulatory environment. This panel will look at the experiences of three academic medical centers in using EHRs for recruitment and the governance structures they are creating to make the process work effectively. Topics will include working with IRBs; managing communications with patients both in terms of content and volume; managing patient preferences for “opting out” of being contacted for research; balancing clinical and research missions and needs; and balancing institutional and individual patient priorities. 

10:30 a.m. – 12:00 p.m.

S28: Panel – Impacts of CDISC Standards on Drug Development Tools and a Learning Health System
L. Becnel, CDISC/Baylor College of Medicine; S. Volchenboum, University of Chicago; S. Arneric, C–Path; B. Delaney, Imperial College of London; A. Nordo, Duke University

The FDA Commissioner recently called for increased use of electronic health record systems’ (EHRs) data to support clinical research and improve safety signal detection at lower cost and greater efficiency as part of a Learning Health System (LHS). These systems often do not share semantics. The consequence of incompatible and non-interoperable platforms is a pervasively inefficient and error-prone system that stifles innovation, leading to increased costs and compromised safety. The Clinical Data Interchange Standards Consortium (CDISC) is a nonprofit standards developing organization that creates global standards for clinical and translational research as part of an LHS. CDISC is poised to connect the “parallel universes” of healthcare systems, academic researchers, and biopharmaceutical/industry where researchers use different vocabularies and data exchange technologies. As of December 2016, FDA requires CDISC standards for new regulated trials’ electronic submissions. Consistent application of these standards to both regulated and unregulated research utilizing data from EHRs helps ensure efficient, meaningful data aggregation and meta-analyses. Here, we will describe CDISC standards for clinical and translational research, the new FDA requirements for their use, and how they have been utilized within different consortia and organizations to propel both regulated and non-regulated clinical research supporting a LHS.

S29: Panel – Data Fitness-for-Use: A Method to Maximize Utility of Poor-quality Data?
J. Klann, Harvard Medical School/Massachusetts General Hospital/Partners Healthcare System; D. Meeker, University of Southern California Keck School of Medicine; M. Rosenman, Northwestern University/Laurie Children's Hospital; S. Murphy, Massachusetts General Hospital/Partners Healthcare System/Harvard Medical School

The Patient Centered Outcomes Research Institute (PCORI) has invested $90 million into PCORnet, a reimagining of the country’s clinical research infrastructure. Multi-state networks of clinical data analytics platforms across all 50 states will perform large-scale comparative effectiveness research. This large-scale reuse of Electronic Health Record (EHR) data magnifies the data quality problems therein. Thus far the data-quality effort has focused on evaluating data sets using global-quality measures and rejecting the entire dataset if it is not compliant. This fails to deal with the granularity of data required by any particular research study. Fundamentally, data with global quality problems (e.g., missing procedures or erroneous lab values) can still be useful for some purposes (e.g., diagnosis-based cohort finding). The panelists each represent a network within PCORnet, and they will present their diverse perspectives on data availability within their unique networks and the steps they are taking to improve their data. Further, each panelist will explore alternatives to the global-quality approach with an eye to needs-based, fitness-for-use approaches tailored to specific studies and use-cases.

Thursday, March 30, 2017

8:30 a.m. – 10:00 a.m.

S35: Panel – Maturity Models for Research IT and Informatics – Reports from the Field   
B. Knosp, University of Iowa Carver College of Medicine; W. Barnett, Indiana University; P. Embi, Regenstrief Institute, Inc.; N. Anderson, University of California, Davis

Understanding the issues involved in planning an institution’s investment in IT and informatics to support research at academic health centers is a challenging endeavor. The level and manner of this planning varies from institution to institution resulting in inconsistent strategies for implementation across academic medicine (for what arguably are similar needs) and unclear linkages between institutional goals and IT investments. To help institutions develop strategies for their investments in IT and informatics that are measurable and align with research agendas, we present maturity indices that can be used as a catalyst for internal discussions and as a tool for establishing standard metrics across academic medicine. Panelists will describe development of these indices and pilot applications in a variety of settings as well as engage attendees on potential next steps in the development and application of these indices.

S36: Panel – Integrative Health Informatics: The VHA Experience
Q. Zeng, George Washington University; J. Goulet, West Haven VA; N. Marshall, Palo Alto VA; C. Morioka, Greater Los Angeles VA

In past two decades, the practice of integrative health has been embraced by healthcare systems and providers. Integrative health combines modern western medicine with complementary integrated health (CIH) (formally call CAM) such as herbal therapy and acupuncture. CIH services are widespread and increasing in the U.S. The Veteran Health Administration (VHA) has been at the forefront of providing integrative healthcare to patients. According to a 2011 study, 125 out of 141 VHA facilities surveyed provided some type of CIH practices, and there was an increase in the number of CIH modalities offered within VHA. Although CIH is widespread, the benefits of CIH are not well understood. For instance, debate on acupuncture’s effectiveness to relieve pain continues in the literature. The cost-effectiveness of CIH is even less understood. In addition, some CIH therapies may be harmful for patients. The interaction between herbal supplement and prescription drug, for example, could potentially be life-threatening. Informatics technologies are being developed to meet the unique needs of CIH research. This panel will bring together a group of informatics, health services, and clinical researchers, who will share their experience in leveraging information technology in integrative health research to fill the knowledge gap regarding the safety, effectiveness and cost-effectiveness of CIH therapies and transform the integrative health research.

10:30 a.m. – 12:00 p.m.

S40: Panel – Distributed Querying across PCORnet: Early Lessons
L. Curtis, Duke Clinical Research Institute; J. Sturtevant, H. Divan, Harvard Pilgrim Health Care Institute; J. Puro, Oregon Community Health Information Network; K. McTigue, University of Pittsburg Medical Center

The mission of PCORnet is to improve the nation’s capacity to conduct clinical research by creating a large, highly representative, national patient-centered network that supports more efficient clinical trials and observational studies. This will be accomplished both through the establishment and efforts of Clinical-data Driven Research Networks (CDRNs) and Patient-Powered Research Networks (PPRNs). Now in Phase 2, PCORnet includes 13 CDRNs comprised of over 80 institutions such as hospitals, health plans, and practice-based networks (together referred to as “sites” or “DataMarts” (DMs)) that have adopted a common data model to support distributed querying across the network. The DMs work in collaboration with the PCORnet Distributed Research Network Operations Center (DRN OC) to enable distributed research querying and related activities to support clinical research. The learning objectives of this panel will include early lessons learned regarding: a) the query fulfillment process overseen by the DRN OC; b) the local DM query fulfillment and governance process for DMs to respond to query requests; c) the process of developing preparatory-to-research query requests through the PCORnet Front Door; and d) the process for implementing research studies across over 40+ DMs within the PCORnet distributed network.

1:30 p.m. – 3:00 p.m.

S44: Panel – Harmonizing Outcome Measures to Increase the Utility of Patient Registries: A Case Study in Atrial Fibrillation
E. Berliner, Agency for Healthcare Research and Quality; R. Gliklich, OM1, Inc.

Patient registries can provide valuable, real-world evidence on the effectiveness, safety, and value of products and interventions to inform decision-making. A particular strength of registries is their ability to collect information on outcomes that are important to patients, providers, and other decision-makers. However, significant variation exists in both the types and definition of outcome measures used in registries, even within specific clinical areas, thereby reducing the utility of registries. The Agency for Healthcare Research and Quality (AHRQ) is attempting to address this variation through the Outcome Measures Framework (OMF), a content model for developing harmonized outcome measures in specific disease areas. AHRQ is assessing the feasibility of using the OMF to develop standardized libraries of definitions in several condition areas. The purpose of this panel is to describe the need for harmonization of outcome measures across the spectrum of clinical care and reporting, share the vision for applying the OMF to specific clinical conditions, and solicit general feedback as well as specific input related to atrial fibrillation.