- Alternative Careers for Biomedical Informatics PhDs
- Beyond the Hype: Developing, Implementing, and Sharing Pharmacogenomic Clinical Decision Support
- How can bio-ontologies support clinical and translational science?
- tranSMART: An Open Source and Community-Driven Informatics and Data Sharing Platform for Clinical and Translational Research
Jessica D. Tenenbaum, PhD,, Duke Translational Medicine Institute, Durham, NC; Marco Sorani, PhD,, Genentech, South San Francisco, CA; Monya Maker, EdM,, Nature Publishing Group, San Francisco, CA; Yael Garten, PhD, Msc,, LinkedIn, Mountain View, CA; Andrew Torrance, PhD, JD,, University of Kansas School of Law, Lawrence, KS.
The number of doctoral training programs in informatics increases every year, however not every doctoral candidate wishes to pursue a traditional career in academia. In addition, the knowledge and skills acquired through scientific training at the doctoral level can be valuable, even critical, for a number of career paths outside of academic research and teaching. This panel will present a diverse set of alternative career paths for which graduates of Informatics programs would be well suited, including patent law, research in industry, academic administration, and scientific journalism. Panelists will describe their own respective backgrounds and career paths, a day in the life in their current position, and how their training prepared them for their jobs. They will also touch on insights gained and lessons learned in exploring the professional landscape through non-traditional paths.
Panel: Beyond the Hype: Developing, Implementing, and Sharing Pharmacogenomic Clinical Decision Support
Presenters: James M. Hoffman, PharmD, MS, St. Jude Children’s Research Hospital, Memphis, TN; Justin Starren, MD, PhD, FACMI, Northwestern University Feinberg School of Medicine, Chicago, IL; Josh F. Peterson, MD, Vanderbilt University School of Medicine, Nashville TN;Mark Hoffman, PhD; Cerner Corporation, Kansas City, MO; Robert R. Freimuth, PhD, Mayo Clinic, Rochester, MN
As pharmacogenomic knowledge is translated into clinical practice, clinical decision support (CDS) in the electronic health record (EHR) will play a crucial role in maximizing the use of pharmacogenomic data over a patient’s lifetime. CDS must be designed to enable clinicians to understand and act upon patient-specific pharmacogenomic data. To accelerate the adoption of pharmacogenomic CDS, the significant effort and lessons learned at innovator sites must be organized and shared among care settings and EHR vendors. While best practices are emerging to represent and share CDS across sites, unique considerations exist for sharing CDS related to genomic data. Panelists from early adopter institutions will compare and contrast their experiences developing and implementing CDS for pharmacogenomics. Perspectives on the integration of pharmacogenomic data into commercial EHRs will be given by a major EHR vendor, and matters specific to knowledge management and sharing CDS for pharmacogenomics in human and machine readable forms will be discussed. Key implementation issues that will be addressed by the panelists include the role of CDS in ordering pharmacogenomics panel tests, the integration of genomic and other clinical data, and the long-term management of genomic data in the EHR.
Models to incorporate genomic and other “omic” (e.g. proteomic, metabolomics, etc.) data into the Electronic Health Records (EHRs) of the future are often discussed. While fundamental changes in EHRs will likely be required to maximize the use of genomic data, pharmacogenomic data are increasingly being used in clinical practice and incorporated into EHRs. Early adopters of pharmacogenomics in clinical practice have recognized that developing and implementing clinical decision support (CDS) for pharmacogenomics is a crucial ingredient for the successful translation of that knowledge into routine clinical practice.
Several pioneering programs to integrate pharmacogenomics into clinical care promote preemptive testing, where genomic data are collected and documented in the EHR early in care before a medication is prescribed. Active CDS is employed at the point of care to remind prescribers to modify drug therapy based on genomic data, which may have been collected in the remote past. Even if genotyping is performed concurrently with a prescription, pharmacogenomic data has lifetime value as new medications are used, and appropriate CDS will enable the pharmacogenomic data to be leveraged each time a relevant medication is used. Methods to readily share CDS for pharmacogenomics across organizations and EHR platforms will be needed to accelerate the implementation of pharmacogenomics into routine patient care.
CDS for pharmacogenomics is emerging within the larger context of the development and implementation of CDS. For example, there is substantial effort being invested to develop, implement, and share CDS to support the care of patients with chronic conditions, such as diabetes and coronary artery disease. However, pharmacogenomic CDS has some unique characteristics compared to other CDS, which must be carefully considered to successfully facilitate sharing CDS for pharmacogenomics.
The objectives of this panel will be to:
- Compare and contrast two early adopters approach to develop and implement CDS for pharmacogenomics. (St. Jude Children’s Research Hospital and Vanderbilt University)
- Review the role of the EHR and clinical decision support in facilitating the testing a panel of drug metabolism genes among appropriate patient populations.
- Describe the practices at each site to store large sets of genomic data and promote it to the EHR.
- Review the use of active and passive CDS for pharmacogenomics at innovator sites, including drug/gene pairs in use and frequency of use.
- Identify challenges and barriers to developing and implementing CDS for pharmacogenomics. Describe strategies for sharing CDS across care settings and EHR platforms.
- Identify unique issues for sharing CDS for pharmacogenomics and methods to overcome these barriers.
Key resources to implement CDS for pharmacogenomics will be highlighted throughout the panel. Examples include PharmGKB and the Clinical Pharmacogenetics Implementation Consortium (CPIC) guidelines, which is a Pharmacogenomics Research Network (PGRN) and PharmGKB collaboration.
This panel was planned in coordination with AMIA genomics working group, and it builds upon this working group’s ongoing series of webinars focused on implementing genomics into the EHR.
Organizer: William R. Hogan, MD, MS, University of Arkansas for Medical Sciences, Little Rock, AR
Participants: Nigam Shah, PhD, MBBS, Stanford University, Stanford, CA); Warren Kibbe, PhD, BS, Northwestern University, Chicago, IL; Melissa Haendel PhD, BA, Oregon Health & Science University, Portland, OR; Mathias Brochhausen PhD, MA University of Arkansas for Medical Sciences, Little Rock, AR
In this panel we will present state-of-the-art applications of ontology to translational science, and derive from that experience what we can learn regarding principles for the future development and use of ontologies to promote translational science. Towards that end, panelists will describe existing benefits of cutting-edge, ontology-driven computing for translational science as well as the pressing issues that require resolution, and how these early results inform the principles we might follow moving forward. We anticipate that a diverse audience will benefit from the discussion: informaticists and researchers will learn the value that ontologies can provide to their research as well as how to choose and integrate them into their work; system developers will learn how to leverage ontologies in support of scientific research; and ontology developers will have a chance to hear and participate in the discussion of proposals for principles for ontology development and use. The intended audience for this panel is therefore biomedical researchers, informatics researchers, ontology researchers, ontology users, health sciences and informatics students, and health information and knowledge management professionals, especially from CTSA sites.
A general description of the panel and issue(s) that will be examined
The panel comprises four experts who are creating and using ontologies as tools to improve the quality and efficiency of translational science plus a moderator. The panelists will discuss their work and what they have learned from it about (1) the benefits of using ontologies, (2) the best ways to integrate ontologies into informatics systems that support translational science, (3) what barriers remain to the use of ontologies, and (4) how to address these barriers going forward. The presentations range from pharmacovigilance to assertions of diagnosis based on EHR data to ontology-based inferences across multiple medically relevant disciplines to solutions for gaps and overlaps in multi-granular ontologies that span basic science, translational science, clinical science, and clinical practice. The panel will help to harmonize the development and use of ontologies within the CTSA framework and provide a forum to discuss issues arising from modularization of ontologies across such a large research network.
Drug safety surveillance using ontologies and free-text clinical notes
As adoption of electronic health records increases, we present an ontology-driven approach for analyzing the unstructured clinical notes, which can enable rapid pharmacovigilance. We show that it is possible to investigate adverse drug event associations with high accuracy (72% sensitivity, 83% specificity) by analyzing textual notes in a clinical data warehouse. We examine suspected associations for confounding via stratification and propensity score matching, performed using ontology-derived features. We find that such an analysis of textual clinical notes could detect adverse drug events 2 years before the official alert. We argue that bio-ontologies may enable data-mining of unstructured clinical notes to expand meaningful use of the electronic health records for post-marketing drug surveillance and for rapid retrospective analysis of adverse event risk elucidated by experimental methods. As an example, using our approach, we found that proton pump inhibitors (PPIs) as a class appear strongly associated with major adverse cardiovascular events, increasing the risk of myocardial infarction by 20-50% depending upon the individual PPI. The association of PPIs with such events was hypothesized based on experimental results that show that PPIs, as a class, elevate plasma levels of asymmetric dimethylarginine, a disease marker and an independent predictor of major adverse cardiovascular events. We will conclude with a brief summary of other translational analyses enabled by bio-ontologies.
Applying ontologies to automate disease-based eligibility criteria
Ontologies, most notably the Gene Ontology (GO), has been highly useful for collecting, organizing, and applying knowledge from multiple knowledge domains, functional domains, and application perspectives. In the GO, the classification of gene production function, biological process, and cellular localization are organized so that the ‘path to the top’ is always true, across all known organisms. This has enabled biomedical researchers investigating gene function to make inference statements based on statements in one organism with homologs in another organism. As an example, published evidence demonstrating that a gene encoding a GTPase-binding protein is involved in ER to Golgi vesicle transplant can be succinctly and precisely annotated using GO and GO evidence codes. Following a similar path, we have used the Disease Ontology, coupled with evidence found in OMIM and the Human Phenotype Ontology, to provide semi-automated methods for assigning disease annotations to patients based on semantically mining EHR data. We have prototyped using those annotations to automate disease-based eligibility registry/cohort selection.
Leveraging ontologies for research reproducibility, resource sharing, and researcher networking
The scientific literature is our collective knowledge about science. However, scientists today also consult and contribute to a large landscape of public resources that include databases for model organisms, reagents, genes, pathways, etc. However, such systems have enormous difficulty in linking their resources to the published literature, as their mentions therein are insufficiently specified. This leads not only to a lack of ability to locate relevant resources, but also to a lack of scientific reproducibility. Ontologies provide a mechanism by which we can logically connect what goes on in the lab or clinic to what is the literature. This includes not only specific definition of research resources by leveraging unique identifiers, but also their logical connection to enable disparate data integration. One barrier to adoption are mechanisms by which to use standardized vocabularies or ontologies during the process of contribution to public databases and the published literature. Ontology-driven tools to support contribution of research resources to public repositories, their identification and contribution to the literature, and researcher networking are being developed to address this issue. The ability to leverage research resources and model systems for translational research relies on one's ability to navigate and reproduce the exponentially growing body of knowledge - ontologies are one mechanism by which we can support these efforts and better understand what we know about science.
Ontological representations for the biobanking domain
We present a strategy to enrich biobank data in a data management environment using i2b2. Our methodology uses an OWL2 implementation of the Minimum Data Set for Biobank Data Sharing (MIABIS). The latter was created and is maintained by the European Biobanking and Biomolecular Resources Research Infrastructure (BBMRI). The ontological representation of MIABIS, called OMIABIS, is based on pre-existing ontologies and follows best practice for biomedical ontologies as formulated by the OBO Foundry.
Due to the management, the operations, and the data collected in the biobanks are distinct, it is a challenge to manually map all of the biobanks’ data into a single i2b2 instance. Currently the biobanks use caTissue. Despite using a singlesoftware application, integration of data is not guaranteed because each biobank creates its own specimen annotation forms with different data elements. We will demonstrate how incorporating OMIABIS into caTissue’s annotation forms ensures integration of hetergeneous, biobanks and the biobank administration data models.
tranSMART: An Open Source and Community-Driven Informatics and Data Sharing Platform for Clinical and Translational Research
Authors: B. Athey, University of Michigan; M. Braxenthaler, Pistoia Alliance; M. Haas, One Mind for Research; Y. Guo, Imperial College London
tranSMART is an emerging global open source public private partnership community developing a comprehensive informatics-based analysis and data-sharing cloud platform for clinical and translational research. The tranSMART consortium includes pharmaceutical and other companies, not-for-profits, academic entities, patient advocacy groups, and government stakeholders. The tranSMART value proposition relies on the concept that the global community of users, developers, and stakeholders are the best source of innovation for applications and for useful data. Continued development and use of the tranSMART platform will create a means to enable “precompetitive” data sharing broadly, saving money and, potentially accelerating research translation to cures. Significant transformative effects of tranSMART includes 1) allowing for all its user community to benefit from experts globally, 2) capturing the best of innovation in analytic tools, 3) a growing ‘big data’ resource, 4) convergent standards, and 5) new informatics-enabled translational science in the pharma, academic, and not-for profit sectors.