3:30 p.m. – 5:00 p.m.
S01: Panel – Building Organizational Capability for Translating Knowledge into Practice: Using FAIR Principles in Diverse Ways
D. Dorr, Oregon Health & Science University; A. Wilcox, University of Washington; G. Melton, University of Minnesota; P. Embi, Regenstrief Institute
With advancing informatics capabilities and rapidly expanding research knowledge, incorporating novel approaches into practice can be daunting. This panel of four diverse informatics leaders will discuss how they help to build organizational capacity to translate knowledge into practice through descriptions of their roles and strategies. The panel will focus efforts on enhancing the findability, accessibility, interoperability, and reuse (the FAIR principles) of data, tools, and standards in their organizations.
10:30 a.m. – 12:00 p.m.
S06: Panel - Governance of the Pluripotent Research Database
H. Lehmann, Johns Hopkins; W. Hersh, Oregon Health & Science University; J. Obeid, Medical University of South Carolina; M. Singleton, Johns Hopkins School of Medicine; A. Solomonides, NorthShore University HealthSystem; U. Tachinardi, University of Wisconsin-Madison
Academic institutions are building large extracts of their electronic health record (EHR) for discoveries, data sharing, registries, multiple machine learning (ML), and other research projects. Many institutional review boards (IRB) are unclear how to provide governance for such pluripotent databases. IRBs are conflicted between the need to protect patients and conform to regulations, through release of the “minimal” amount of information possible, while at the same time promoting research. Using a real case, an example of an IRB Big Data request submitted to the Johns Hopkins Data Trust Research Subcouncil will be presented. Four panelists from the informatics community will provide five to ten minute comments of their perspectives. A fifth panelist, the assistant dean responsible for human subject protection at Johns Hopkins, will respond. We will then have a brief response by the panelists and an open discussion by the audience.
S08: Panel – Can Informatics Return the Joy of Medicine? A Real World Pilot in Neurology
J. Liberman, J. Jones, J. Cooper, W. Stewart, Sutter Health
Clinicians strive to deliver high quality and personalized care. While precision care is theoretically possible given advances in knowledge and the availability of information, it is practically impossible to deliver because of limitations in the technology in place today. Through a collaboration with healthcare innovators, developers, researchers, clinicians, and patients, Sutter Health is designing, developing, and testing (in real-world clinical practice) a unique point-of-care application called neuroSHARE. neuroSHARE is designed to address the challenges of delivering precise care by automating the curation, organization, and visualization of relevant data and knowledge in an easy-to-use, point-of-care format. Participants will learn: practical, real-world challenges of delivering informatics to the clinician at the point-of-care; how to design an application to optimize adoption by clinicians; methods for rapid-cycle, test-and-learn experimentation in clinical practice; how clinicians, in a busy practice setting, think about and prioritize adoption of digital health innovations.
1:30 p.m. – 3:00 p.m.
S11: Panel – The CTSA Program Center for Data to Health
M. Haendel, Oregon Health Sciences University; K. Holmes, Northwestern University; S. Mooney, University of Washington; C. Chute, Johns Hopkins University; J. Guinney, Sage Bionetworks
The new Center for Data to Health (CD2H) welcomes the entire translational informatics community to this panel discussion to meet the CD2H team and hear about the plans for collaborative translational informatics - built on the structure and vision of the Clinical and Translational Awards Program. The session provides an opportunity to discuss ideas for how to best leverage and engage AMIA pioneers and activities in open science, clinical and biomedical informatics, and in health science libraries for the purposes of creating a welcoming ecosystem of shared data, tools and algorithms, and expertise across clinical and translational science institutions.
3:30 p.m. – 5:00 p.m.
S16: Panel – The Future of Evidence Synthesis: Incorporating Heterogeneous, Non-traditional, and Patient-specific Evidence towards Evidence-based Precision Medicine
I. Sim, University of California San Francisco; J. Ioannidis, Stanford University; R. Horwitz, Temple University; J. Silverstein, University of Pittsburgh; J. Schneider, University of Illinois at Urbana-Champaign
As clinical practice becomes more focused on personalized and precision medicine, there are increasing challenges for evidence-based practice, which combines the best research evidence, clinical expertise, and patient values and preferences in order to support clinical decision making. Identifying the best research evidence relevant to a particular patient requires new approaches to evidence synthesis. New technologies are emerging for both the current standard (group-oriented Randomized Controlled Trials) and “beyond the RCT” studies.
8:30 a.m. – 10:00 a.m.
S21: Panel – Moving Genomics into the Clinic: Informatic Approaches from eMERGE, ClinGen, HL7 and GA4GH
M. Williams, Geisinger
Genomics is being integrated into clinical care at an accelerated pace as evidence accrues for its value and impact on health outcomes. Major challenges exist that interfere with clinical translation. This includes the lack of standards for variant interpretation, representation, and knowledge management, as well as the availability of information at the point of care to guide clinicians and patients. Two NHGRI-funded projects, the electronic Medical Records and Genomics (eMERGE) network and the Clinical Genome resource (ClinGen), are working in conjunction with other efforts such as Health Level 7 (HL7) and Global Alliance for Genomic Health (GA4GH) to create and test resources that address these challenges. The objectives of this panel are to: present progress on resources to support variant curation (ClinGen), point of care informational resources (eMERGE and ClinGen), and informatic standards (HL7, GA4GH); inform attendees how to utilize these resources; solicit feedback from attendees on how these resources can be improved from the perspective of end-users.
10:30 a.m. – 12:00 p.m.
S26: Panel – Improving Standardization and Interoperability of Common Data Models for Clinical and Translational Research
G. Jiang, Mayo Clinic; J. Duke, Georgia Institute of Technology; D. Meeker, University of Southern California; E. Oliveira, REACHnet; M. Rocca, FDA; H. Solbrig, Mayo Clinic; S. Murphy, Harvard Medical School
Clinical and translational research studies increasingly involve the manipulation of large datasets. The development of integrated data repositories (IDRs) based on common data models (CDMs) are needed to both lower the effort required for, as well as incentivize the process of, standardizing clinical research data in clinical and translational research. In particular, the harmonization of CDMs can make it easier for researchers to share their data and research results across different informatics platforms and to broadly benefit the community. This thematic panel will highlight five projects that are developing technology infrastructure and tools for the CDM harmonization, standardization and applications. Attendees will learn about novel open source CDM-based informatics platform, and how CDMs are used in data sharing platforms. Attendees can engage with panelists on the CDM harmonization and interoperability challenges and potential solutions in support of clinical and translational research applications.
10:30 a.m. – 12:00 p.m.
S27: Panel – Cancer Precision Medicine in Clinical Practice: An Integrated Approach
C. Cole, O. Elemento, Weill Cornell Medicine; D. Artz, Standard Molecular, Inc/Weill Cornell Medicine; S. Malhotra, T. Campion, Weill Cornell Medicine
To fulfill the promise of Precision Medicine, genomic data from the research lab must be integrated into clinical practice. In this panel, experts share their experience of successfully integrating a next-generation sequencing test (EXaCT-1) into the clinical setting. A core guiding principle of the initiative was to store discrete, variant-level genomic data in both the EHR and research database; this allows for customizable data views, decision support for clinical care, and analysis for research purposes. The monumental task required coordination and build between multiple departments, vendors, clinical and research information system personnel. Each panelist relays the unique challenges and pragmatic aspects of project design, information system build, system and workflow integration, data flow and storage from their perspective. We also describe our strategy around use of controlled vocabularies and existing vendor architecture for storing genomic data, so as to prevent future invalidation of data representation that may occur as these systems evolve independently.
1:30 p.m. – 3:00 p.m.
S31: Panel – Towards Large-scale Predictive Drug Safety: A Systems Pharmacology Perspective
P. Zhang, IBM T.J. Watson Research Center; K. Burkhart, FDA; L. Li, Ohio State University; A. Ma'ayan, The Icahn School of Medicine at Mount Sinai; N. Tatonetti, Columbia University
Adverse drug reaction (ADR) is a major burden for patients and healthcare industry. Systems pharmacology, which involves the application of systems biology approaches, combining large-scale experimental studies with computational analytics, can enhance the understanding of ADRs by looking at the effects of a drug in the context of cellular networks as well as exploring relationships between drugs. Recent efforts in high throughput experiments have generated a huge amount of data across the multiple biological scales of the organism, across a wide range of time scales, and across multiple species. These data sets provide unprecedented opportunities for systems pharmacology, but impose great challenges in big data management, mining, and integration. Furthermore, to materialize the true potential and impact of systems pharmacology approaches, 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 informatics and systems pharmacology for drug safety and identify challenges and opportunities. Panel participants will synthesize their perspectives on these key issues and likely future developments in this area, explore a diverse set of topics, and engage in thoughtful discussion with the audience.
8:30 a.m. – 10:00 a.m.
S36: Panel – Nursing Documentation and the Clinical Research Informatics Pipeline
J. Klann, Harvard University School of Medicine/Massachusetts General Hospital/Partners Healthcare; S. Collins, Brigham and Women's Hospital/Harvard University School of Medicine; K. Cato, Columbia University/New York Presbyterian Hospital; R. Waitman, University of Kansas Medical Center; B. Westra, University of Minnesota
Data research repositories and networks, the backbone of many retrospective informatics research projects, contain selected “most relevant” data transformed from the electronic health record. Data models commonly include patient demographic information, encounter context information, and key clinical data (e.g., medication, diagnosis, and procedure). Nursing data, however, is seldom included. Common Data Models (e.g., OMOP and PCORnet CDMs), research networks (PCORnet and NIH ACT), and ontology/terminology curation organizations (e.g., BioPortal) have largely ignored nursing data. However, in hospitals, nurses provide most direct care to patients, diligently document patient states, and routinely make decisions and implement interventions based on these clinical assessments. Flowsheet data provide structured documentation that goes beyond physiologic measures to assessments of well-being and wellness (e.g., activities of daily living), levels of risk, and responses to treatment. Nursing notes offer additional context, and they can be used to predict morbidity and mortality. Here, we explore several exciting endeavors at several health systems to bring nursing documentation into the pipeline of clinical research informatics, and how it is being applied to healthcare problems. In this time of heavy investment into data research networks, this panel addresses the timely need to expand our view of what constitutes core clinical data.
10:30 a.m. – 12:00 p.m.
S40: Panel – Deep Learning for Healthcare - Hype or the Real Thing?
J. Sun, Georgia Institute of Technology; B. Westover, Massachusetts General Hospital; H. Yu, University of Massachusetts; D. Sontag, Massachusetts Institute of Technology; M. Ghassemi, Massachusetts Institute of Technology/Verily
Deep learning methods refer to variants of neural network models where two or more layers of nodes are integrated with each other. Use of these models is in a renaissance given the substantial advances in the accuracy of neural network models for computer vision, natural language processing and speech recognition. Availability of large scale datasets, advances in high-performance computational devices such as Graphics Processing Units (GPU), and developments of new optimization techniques have permitted efficient learning of deep neural networks for a diversity of application domains. Deep learning methods are being applied in the healthcare domain: Automated feature learning: Autoencoders have been used for phenotyping purposes using time-series data. Skip-gram methods were used for learning the representation vectors for medical codes such as diagnosis codes and medication codes. Recurrent networks (RNN) have been used to learn representations from clinical notes. Convolutional networks (ConvNet) also showed great potential for learning effective representations of continuous vital signs. Accurate predictive models: Training a high-capacity model for accurate prediction is one of the most sought-after direction from many researchers as well as practitioners. ConvNets have shown state-of-the-art performance for detecting retinopathy from retinal images and detecting skin cancer from skin photographs. RNN has also shown great potential for multi-label diagnoses prediction and early detection of heart failure onset. Synthetic data generation: Generating synthetic electronic health records is another popular direction due to the sensitive nature of private records. Generative adversarial networks are actively being used to generate synthetic claims data and lab measures, or even to guarantee some level of privacy. However, deep learning models still have many limitations in support healthcare applications: Interpretation: The deep learning models are often complex black boxes, which are difficult to understand. However, clinical practitioners often prefer and trust simpler and interpretable models. We will discuss how to bridge the gap between model complexity and interpretation. Causal learning: Great predictive models do not necessarily lead to causal discovery. For clinical applications such as treatment selection, we need to know what treatment change will lead to the best outcome, which often requires identifying causal relations.
1:30 p.m. – 3:00 p.m.
S44: Panel – Developing Data Management Infrastructure Supporting Innovation in Multi-center Clinical Trials at-scale: A Report from the Trial Innovation Network
D. Gabriel, Duke Clinical Research Institute; B. LaSalle, University of Utah; B. McCourt, Duke Clinical Research Institute; P. Harris, Vanderbilt University
The Trial Innovation Network (TIN) is a collaborative, national multidisciplinary platform that focuses on operational innovation, excellence and collaboration to support multi-center clinical trials, utilizing the resources and expertise of awardees of the National Center for Advancing Translational Science (NCATS) Clinical and Translational Science Awards (CTSAs). The panel consists of representatives from the three Trial Innovation Centers (TICs) Data Management and Harmonization Working Group (DMHWG) and the Recruitment Innovation Center (RIC). Results of data management innovation and activities undertaken by the group are discussed.