December 2, 2016 - Joshua Lee, MD, Associate professor, Loyola University Health.
"To test or not to test? That is the question?" Using the noble power of informatics to ensure appropriateness in diagnostic testing.
Abstract:Given the increasing scrutiny on the effectiveness of health care expenditures, we as informaticists are called upon to leverage tools to scrutinize this spend. One significant cost in care across the continuum is diagnostic testing. In this presentation, I will focus on historic efforts to direct providers towards more effective clinical lab ordering as well as the evolution of this clinical decision support paradigm. Further, given our nationwide efforts towards the reduction, if not elimination, of hospital acquired infections, I will focus on the ways in which targeted interventions can improve national quality measures related to these infectious risks.
Bio: Dr. Joshua Lee joined Loyola University Health as the Chief Health Information Officer in January 2016. . He oversees all informatics activities and also coordinates educational and research efforts with the Stritch School of Medicine of the Loyola University of Chicago. Prior to Loyola, Dr. Lee served as Chief Information Officer for Keck Medicine of USC from 2012-2016. He served as the associate director for informatics for the Southern California-Clinical Translational Science Institute. At USC, he was responsible for the enterprise wide deployment of Cerner Millennium. From 2003-2012, he was the Chief Medical Information Officer at UC San Diego Health. There, he was responsible for the enterprise wide deployment of a continuum of care health record (Epic Hyperspace) and participated in the attainment of HIMSS analytics Stage 7 for the health system. Dr. Lee earned his MD degree at the University of California, San Francisco, and completed his residency and chief residency in internal medicine residency at Brigham and Women’s Hospital.
November 18, 2016 - Claudiu Farcas, Ph.D. Assistant Research Scientist at DBMI, UCSD.
Scaling Science through Software Engineering
Abstract:The days of pencil and paper discoveries are long past. From novel algorithms and analysis methods to artificial intelligence explorations, software is now THE source of innovation and deeply embedded into the academic expectation. However, outside of the computer science realm, most researchers face significant challenges designing and developing software at scale. Here we learn about the importance of software engineering and its role towards quality products and robust development practices.
Bio: Claudiu Farcas, PhD is an assistant research scientist at Calit2 and DBMI. His research interests are software engineering for embedded and large-scale distributed systems, virtual execution environments, programming languages, networking, and Web technologies. He currently leads on the technical side multiple large projects at DBMI focusing on patient privacy, secure human-subjects data exchanges, distributed queries, and biomedical data indexing.
November 4, 2016 - Wei Wei, Ph.D. student, DBMI, UCSD.
Convolutional neural network: a good feature engineer in the Medical Subject Headings (MeSH) assignment task
Abstract:Assigning Medical Subject Headings (MeSH) is an important step in indexing biomedical research articles in the National Library of Medicine (NLM). Automated MeSH assignment methods have been developed to help NLM indexers, and most of these methods use manually crafted features. In recent years, data-driven features learned using deep learning methods have been proven to capture abundant information, and therefore have been widely used in natural language processing (NLP) research. We studied the distributed representation features generated from a deep learning method, Convolutional Neural Networks (CNN) in a MeSH assignment task, and found that they can result in better assignment performance when compared to various types of manually crafted features, provided that enough training data are available. In addition, these data-driven features can improve model performance without requiring additional training data.
Bio: Wei Wei is a PhD student in the Department of Biomedical Informatics, and the Bioinformatics and Systems Biology program. Wei works on automatic assignment of MeSH terms to the biomedical literature, and methods that improves dataset retrieval performance. Wei received his Master’s degree in Biomedical Informatics from the University of Pittsburgh, and the Bachelor’s degree in Biology from Zhejiang University, China.
November 4, 2016 - Henry (Yingxiang) Huang NLM Ph.D. fellow, DBMI, UCSD.
A Global Context-aware Medical Events Prediction System
Abstract: Currently, there is a lack of efficient methods for hospitals to share information, so diagnostic prediction models are often only built off of data from one hospital site. Isolated models would benefit from sharing information between hospitals to build more global and general models instead of ones that could potentially be biased because of the data. However, because of privacy concerns, we can’t readily transfer patient information between hospitals and simply train one model with the combined patient data. Building upon previous work where medical events (diagnoses, lab tests, prescription, and etc) from a corpus was represented as continuous vectors, I applied fusion techniques on these medical events vectors from two different hospital sites, aligning them to the same vector space. This talk will assess the performance of space alignment techniques such as Procrustes, metric and non-metric multidimensional scaling (NMDS), and present an improved version of MDS.
Bio: I am a second year Ph.D student in the Department of Biomedical Informatics at UCSD. My current research focus is on building diagnostic prediction models to provide precision medicine at the individual level. Such models can ideally be used by both physicians to visualize possible future events in their patients, and patients to understand their individualized versions of their diagnoses. I got my B.A. in Molecular Cellular Biology and B.A. in Integrative Biology from UC Berkeley.
October 28, 2016 - Cheryl Brown, Ph.D., Associate Professor, Department of Political Science and Public
Administration, The University of North Carolina at Charlotte.
Trust, Education, and Culture: Big Data Pre-Steps beyond Privacy and Ethics for Precision Medicine
Abstract:Big data’s development from Laney’s (Gartner) 3V’s framework of volume, velocity, and variety has expanded to four parameters including veracity, and in some domains, as many as six V’s in adding value and variability to this ubiquitous concept. The translation of big data to knowledge (BD2K) for precision medicine, however, has significant pre-steps with societal impact. This talk will critically explore the complex web of trust, education, and culture as preliminary factors in enlarging the volume and other V’s affecting precision medicine and outcomes. Unraveling the web for close examination is crucial for increasing the population base of participants for precision care.
Bio: Dr. Cheryl L. Brown is an Associate Professor in the Department of Political Science and Public Administration at the University of North Carolina at Charlotte. Her current research focuses on privacy, ethics, and trust in big data for precision medicine and health quantification for connected and autonomous vehicles, both projects she presented at National Institutes of Health conferences. She teaches courses on privacy, ethics, and governance of big data; cybersecurity policy; legal and policy issues of the Internet of Things; Chinese domestic and foreign policy; and East Asian foreign policy. Dr. Brown received her Ph.D. and M.A. degrees in Political Science, specializing in Chinese studies, from the University of Michigan and B.A. degree in Political Science from the University of Florida. She held a visiting scholar position at the East-West Center in Honolulu, Hawaii and participated in the Summer Cyberlaw Program at the Berkman Center of Harvard Law School.
October 21, 2016 - Mona Vij, Staff Research Scientist, Intel Corporation.
Software Guard Extensions (SGX) and applications in cloud computing
Abstract:Software systems are responsible for protecting the confidentiality and integrity of many types of secrets on today’s open platforms. The challenge is that we typically rely on privilege separation techniques to protect numerous software layers with a very large code base. The likely risks of flaws can lead to privilege escalation and compromise the security of the entire system. In addition to software adversaries, there is also desire to protect against bad firmware, rogue devices and hardware adversaries, leading security community towards building Trusted Execution Environments (TEE) with a hardware root of trust. Unfortunately, many of the existing TEEs are vulnerable to hardware attacks and/or require lots of software to be re-written to support and integrate them. Intel’s new technology Secure Guard Extensions (SGX) provides a great leap forward this. On one hand, it strengthens security, especially against hardware attacker; on the other hand, it provides scalable security within mainstream environments by enabling software developers to develop and deploy secure applications on open platforms with a minimal trusted code base. We present the current SGX architecture and discuss some of the challenges with such a solution in cloud environment. We wrap up the talk with couple of SGX cloud usages on privacy preserving analytics and Block chains.
Bio: Mona Vij is a Research Scientist and leads the Cloud and Data Center security team at Security and Privacy Research Lab in Intel Labs Hillsboro, Oregon. In addition she is an SGX architect and leads the university engagement with a large number of universities in the area of cloud security with a focus on SGX. She has extensive research and development experience in trusted computing, virtualization, device drivers and has been a security and operating systems researcher for over 20 years. She has a Masters in Computer Science from the University of Delhi, India and a Bachelor of Science in Mathematics from St Stephen’s College, Delhi.
October 14, 2016 - Ferdinand Velasco, MD, Chief Health Information Officer, Texas Health Resources
Hard-wiring reliable care: a systematic approach for designing,
deploying and sustaining evidence-based care models to achieve better,
safer patient care
Abstract:As part of a strategic initiative to improve the reliability of its care delivery, in 2015 Texas Health Resources launched two parallel efforts:
1. Building a culture of high reliability through the adoption of high reliability principles, the promotion of professional standards of performance, and the deployment of error prevention tools.
2. Instituting reliable care blueprinting (RCB), a formal methodology for designing care models based on clinical evidence, hard-wiring these models using information technology, and monitoring clinical performance through advanced analytics.
In this presentation, we discuss our early experience with the latter of these two initiatives (RCB). Specifically, we will provide an overview of the RCB approach, our organizational structure in support of RCB development and deployment, lessons learned from selected case studies, and early results and insights.
Bio: As Texas Health’s chief health information officer, Dr. Ferdinand Velasco leads the system’s clinical decision support, medical and nursing informatics, business intelligence and data analytics functions across the continuum of care. Through his and his team's leadership of the implementation of the electronic health record, Texas Health was awarded the Enterprise Davies Award for Excellence in 2013.
Prior to joining Texas Health Resources in 2002, Dr. Velasco served as an assistant professor and physician champion for the implementation of computerized physician order entry (CPOE) at the Weill Medical College of Cornell University. He simultaneously practiced as a cardiothoracic surgeon at NewYork-Presbyterian Hospital.
A Fellow of the Healthcare Information Management Systems Society (HIMSS), Dr. Velasco serves on the North America Board of HIMSS and previously chaired the society’s Quality, Cost, Safety Committee. Modern Healthcare honored Dr. Velasco as an inaugural member of the Top 25 Clinical Informaticists in Healthcare. He received his medical degree from the University of California, Los Angeles School of Medicine.
October 7, 2016 - Jina Huh, Ph.D., Assistant Professor, DBMI, UCSD.
Visual social media for health
Abstract: In this talk, I will present an ongoing study to understand how visual social media is used for health related research. I will show how different forms of research is being published in PubMed vs. ACM digital library archives about visual social media platforms (e.g., Flickr, Pinterest, Snapchat). This presentation will be interactive, soliciting feedback and analysis from the students on the keyword search results from ACM and PubMed.
Bio: Jina Huh is Assistant Professor in the Department of Biomedical Informatics at the University of California San Diego. She studies mobile health applications and social media to improve daily health management. She is a PI on an NIH grant to improve information quality on online health communities and an NSF grant to improve Latino family routines through visual social awareness program based on phone-based acoustic sensors. She was an Assistant Professor at the Department of Media and Information at Michigan State University. She received the NLM postdoctoral fellowship at the University of Washington, a PhD from the University of Michigan School of Information, a Masters in HCI from Carnegie Mellon University, and a BA from Korea National University of Arts.
September 30, 2016 - Ricky Bloomfield, MD, Assistant Professor of Pediatrics, School of Medicine, Duke University.
Innovation In and Around the EHR
Abstract:Dr. Bloomfield will discuss a high-level overview of technology innovation at Duke Health. This will include patient-centric tools such as Apple’s HealthKit and a custom MyChart mobile app as well as provider-centric technologies such as SMART on FHIR. He will also discuss the creation of the first ResearchKit app designed to study children, which has now expanded to Africa.
Bio: Dr. Bloomfield joined Duke in 2013 to oversee mobile technology initiatives within the healthcare system. In addition to rolling out mobile applications associated with the Epic EHR, he is collaborating with multiple entities both inside and outside Duke to establish an open technology platform for mobile health innovation. During his residency, Dr. Bloomfield founded a successful software company creating health and social networking apps for iOS devices that has enjoyed over 15 million downloads to date.
Dr. Bloomfield led the integration of Apple HealthKit at Duke in August 2014 (the first health system to do so) and is helping facilitate its use among interested faculty for both clinical and research purposes. He also led Duke to become the first health system with the Epic EHR to incorporate the SMART on FHIR platform, enabling a new generation of plug and play medical apps. In addition, he is a Co-PI on Duke’s first ResearchKit app, Autism & Beyond, and also provides guidance to other key initiatives across the health system, including telemedicine, a health accelerator, secure messaging, custom apps, and clinical decision support. He still practices clinically as a Pediatric and Internal Medicine hospitalist.
Prior to joining Duke, Dr. Bloomfield completed his Internal Medicine and Pediatrics residency at UNC Hospitals in Chapel Hill, NC, followed by a Pediatric Chief Resident year. A native of eastern North Carolina, he’s never found a good enough reason to permanently leave his home state. In his spare time, he enjoys playing jazz saxophone in a local big band, competitive swimming, writing iOS and Android apps, volunteering in a local Hispanic free clinic, and spending time with his wife and two 9 year-old daughters.
You can follow him on Twitter @rickybloomfield or on his blog at www.rickybloomfield.com
September 23, 2016 - Tsung-Ting Kuo, Ph.D. Postdoctoral Fellow, DBMI, UCSD.
ModelChain: Decentralized Privacy-Preserving Healthcare Predictive Modeling Framework on Private Blockchain Networks
Abstract: Cross-institutional healthcare predictive modeling can accelerate research and facilitate quality improvement initiatives, but most existing privacy-protecting methods are based on centralized architecture which presents security and robustness vulnerabilities. We describe a new framework, ModelChain, to adapt Blockchain technology for privacy-preserving machine learning, without revealing any patient health information. We also discuss the benefits and potential issues of applying Blockchain technology to increase interoperability between institutions.
Bio: Dr. Tsung-Ting Kuo is a Postdoctoral Fellow in DBMI, School of Medicine, UCSD. He earned his Ph.D. in National Taiwan University (NTU), M.S. in National Chiao-Tung University (NCTU), and B.S. in National Cheng-Kung University (NCKU), all in Computer Science. He was the major contributor of the UCSD DBMI team, which is one of the winners of the Use of Blockchain in Health IT and Health-related Research Challenge hold by The Office of the National Coordinator for Health Information Technology (ONC) in 2016. He was a runner-up for the Best Dissertation Award from Association of Computational Linguistic in Taiwan (ACLCLP) and Taiwan Institute of Electrical and Electronic Engineering (TIEEE) in 2014. He is also a winner of NTU Outstanding Performance Scholarship in 2011, and many other awards. He was a major contributor in the NTU team, a five-time champion of Association for Computing Machinery (ACM) Knowledge Discovery and Data Mining (KDD) Cup from 2008 to 2013. His current research interests include machine learning, natural language processing, and Blockchain technology in biomedical and healthcare domain.
June 3, 2016 - Adam Landman, MD, Assistant Professor of Medicine (Emergency Medicine), Harvard Medical School.
Accelerating Digital Health Innovation at Academic Medical Centers
Abstract: Digital health tools ranging from apps to analytics have the potential to enable value-based care delivery. Academic medical centers are well positioned to help ensure digital health innovations match clinical workflows and improve quality, safety, and efficiency, while reducing costs. Brigham and Women’s Hospital (BWH) is driving internal culture change and facilitating both internal and external digital health innovation through events, information systems process redesign, and partnerships with industry. We will share our experiences from three years of Hackathons, including the 2015 BWH Open.Epic Challenge to build working applications on top of Epic using Fast Healthcare Interoperability Resources (FHIR) services. We will describe our Digital Health Innovation Group that reduces IT barriers and provides pathways for digital health innovation to move forward from idea to the patient. Finally, we share how we are partnering with early stage external start-ups to accelerate clinical pilots, generate evidence, and allow our patients and staff to benefit from novel solutions.
Bio: Adam Landman, MD, MS, MIS, MHS, is an attending emergency physician and Chief Medical Information Officer for Health Information Innovation and Integration at the Brigham and Women's Hospital in Boston, MA. He is also an Assistant Professor at Harvard Medical School. Dr. Landman has designed and developed mobile apps to securely capture and store clinical images and to improve electronic medication administration and reconciliation (e-MAR). He is currently responsible for clinical oversight of the implementation and integration of a new hospital laboratory system, multiple projects related to a new enterprise EHR implementation, as well as health information innovation programs.
May 27, 2016 - GQ Zhang, Ph.D. Professor of Biomedical Informatics, University of Kentucky.
The role of ontologies in clinical and translational informatic
Abstract: This talk will present the role of ontologies in clinical and translational informatics, illustrated using specific research data resource examples. One such resource is the National Research Resource grant (R24HL114473), to establish the NHLBI National Sleep Research Resource, a comprehensive, easily accessible and well-annotated national repository of sleep data to make data from more than 50,000 sleep studies available to biomedical researchers. The second is the MEDCIS Data Repository for the NINDS-funded Center for SUDEP Research (U01NS090408), a prospective collaborative involving 14 institutions for the understanding of sudden unexpected death in epilepsy. The use of a Sleep Domain Ontology and Epilepsy and Seizure Ontology in both resources are discussed. In the second part of the talk, recent developments in non-lattice auditing of biomedical ontologies such as SNOMED CT and GO are presented.
Bio: Dr. Zhang is Professor of Internal Medicine and Computer Science at the University of Kentucky, where he is the Director of the Institute of Biomedical Informatics; Chief of the Division of Biomedical Informatics (https://bmi.med.uky.edu); Co-Director of Informatics Core, Center for Clinical & Translational Science (http://www.ccts.uky.edu/ccts/BMI_Core); and Director, Informatics & Data Analytics Core, NINDS-CWW Center for SUDEP Research (sudepresearch.org). Prior to joining the University of Kentucky, his role at Case Western Reserve University included Division Chief of Medical Informatics, Co-Director of Biomedical Research Information Management Core of the Case Western CTSA, and Associate Director for Case Comprehensive Cancer Center while performing duties as a tenured professor in the Case School of Engineering. His research interests include data science and bigdata in biomedicine, large-scale, multi-center data integration, biomedical ontology development, information retrieval, query interface design, and agile, interface-driven, access-control grounded software development. These interests are reflected in his over 130 publications in Computer Science and Biomedical Informatics. Over more than a decade, Dr. Zhang has developed a range of clinical research informatics tools for data capturing, data management, cohort discovery, and clinical decision support, such as VISAGE (PMID: 21347154), MEDCIS (PMID: 23686934), OnWARD (PMID: 21924379), OPIC (PMID: 23304354), EpiDEA (PMID: 23304396), and Cloudwave (PMID: 23920671). Supported by multiple federal- and foundation-funded awards and acclimated in a multi-disciplinary team-science, collaborative setting, Dr. Zhang effectively brings cutting-edge computer science and informatics methodology to addressing biomedical data/big data challenges through the translation of theory, algorithms, methods and best practices to functional and usable tools impacting the entire clinical research data lifecycle.
May 20, 2016 - Rebecca A. Marmor, MD, NLM/NIH Clinical Informatics Fellow, DBMI, University of California San Diego.
Online Health Communities: A Novel Data Source for the Study of Breast Cancer
Abstract: Online health communities provide a rich data source to explore patient preferences regarding surgical treatment for breast cancer. We will present a longitudinal analysis of repeated participation in an online health community for breast cancer on positive emotion and discussion of health related topics. We will also present a qualitative analysis of reasons why patients choose contralateral prophylactic mastectomy, as described on an online health community.
Bio: Rebecca Marmor, MD is a General Surgery Resident in the Department of Surgery and Clinical Informatics Fellow at UCSD. Her clinical interests are surgical oncology and research interests are in online health communities, patient education and the impact of social dynamics on patient preferences regarding surgical treatment.
May 13, 2016 - Todd Ferris, MD, Chief Technology Officer, School of Medicine, Stanford University.
EHR Research: Successes & Failures - The Stanford Experience
Abstract: The Stanford Translational Research Integrated Database Environment (STRIDE) was established in 2004 to facilitate the research and educational use of Stanford patient data. In the more than 10 years STRIDE has been in operation, the Stanford Center for Clinical Informatics (SCCI) team has received tens of thousands of consultation requests, has provided several thousand data sets, and has seen hundreds of publications from those data sets. This talk will explore some of the notable publications, both those with positive findings, as well as those with negative or unsuccessful outcomes. Along the way, the various tools and techniques used by the SCCI team will be explored.
Bio: Todd Ferris, MD, MS is the Chief Technology Officer at Stanford University School of Medicine, Chief Informatician for the Stanford Center for Clinical Informatics (SCCI), and Associate Program Director (Didactic Curriculum) for the Stanford Clinical Informatics Fellowship. He is a graduate of Stanford's Biomedical Informatics Training Program and board certified in Clinical Informatics. Previously he was the information security and privacy officer for Stanford's School of Medicine and has an extensive background in privacy and security of medical information. He was an advisor to the California Office of Health Information Integrity as a member of the California Privacy and Security Advisory Board. In his role at Stanford, he has managed several projects involving using EHR data for research and health data collaborations between organizations.
May 6, 2016 - Cenk Sahinalp, Ph.D. Professor of Computer Science, Indiana University.
Compression and compressed computation for high throughput sequencing data
Abstract Massive genomic read data are challenging to store and
transmit. Fortunately, the human genome is highly repetitive, providing
ample opportunity for compression. Short read collections can be
represented raw or in mapped formats. For raw read data we have
developed SCALCE, a compression booster to reorder reads so as to
improve their locality of reference and thus their collective
compressibility through general purpose compression methods. For mapped
read data, we have developed DeeZ, a method to build consensus contigs
on which reads are represented simply as differentially encoded
positions. SCALCE and DeeZ are among several tools that have been
recently developed to reduce the sequencing data footprint on networks
and storage systems. We will discuss the results of a recent study
evaluating all available tools in the literature on a ISO/IEC JTC 1/SC
29 Working Group 11 (a.k.a. MPEG) approved benchmarking data set, for
the purposes of setting up a sequence representation and compression
standard. Although the repetitive nature of the human genome is highly
useful when compressing relevant sequence data, it poses major
challenges in determining exact nature of an individual genome sequenced
through short read technologies. One major problem in this direction is
the resolution of the exact allelic composition of highly polymorphic
genes. For example, the CYP gene family, which is involved in the
metabolism of more than 90% of all clinically prescribed drugs are
highly polymorphic. Reads from the CYP region are difficult to map and
assemble. In particular, CYP2D6, a member of this family, involved in
metabolizing a quarter of all available drugs is particularly important
for pharmacogenomic purposes and its genotyping is highly recommended
prior to treatment decisions. Unfortunately, in order to accurately
interpret short read data involving the CYP2D6 gene, one needs to handle
high sequence similarity and genetic recombinations between CYP2D6 and
the related pseudogenes CYP2D7 and CYP2D8. For addressing these problems
we have develooped CYPIRIPI, the first computational tool to accurately
infer CYP2D6 genotype at basepair resolution. CYPIRIPI can resolve
complex genotypes, including alleles that are the products of
duplication, deletion and fusion events.:
BIo: Dr. S. Cenk Sahinalp is a Professor of Computer Science at Indiana University, Bloomington. He is currently a Simons visiting fellow at U.C. Berkeley, the co-director of the Center for Genomics and Bioinformatics at IU, an associate member of the Vancouver Prostate Centre and a professor at Simon Fraser University (on-leave). His lab develops algorithmic infrastructure for problems in computational genomics and transcriptomics - especially in the context of cancer, computational methods for exploring biomolecular networks and long non-coding RNAs.
April 29, 2016 - Siddhart Singh, MD. Clinical Assistant Professor of Medicine, UCSD.
Build it, and they will come: Harnessing EMR data to build disease- specific cohorts
Abstract: With the proliferation of electronic health records, there is a unique opportunity to create large cohorts of patients with specific diseases. By combining structured and narrative data (from clinical notes), these cohorts are rich treasure troves of information. And once these are built, it opens up unique opportunities and collaborations to answer comparative effectiveness questions, genotype-phenotype questions and improve population health in an unprecedented manner"
Bio: Sid Singh, MD, MS, is an Assistant Professor of Medicine in the Division of Gastroenterology, and Clinical Informatics Fellow at UCSD. His clinical interests are in inflammatory bowel diseases, and research interests are in comparative effectiveness, patient-centered research. He's also involved in knowledge synthesis, assessing quality of evidence, guideline development and implementation at the American Gastroenterological Association.
Apil 22, 2016 - Adam Wright, Ph.D., Associate Professor of Medicine, Brigham and Women's Hospital.
Improving Clinical Decision Support Reliability Using Anomaly Detection Methods
Abstract: Clinical decision support (CDS) systems, when effectively designed and used, are powerful tools for improving the quality and safety of healthcare. However, CDS systems are complex to develop and maintain and, our research group has found, frequently malfunction. A range of causes, including changes in codes and fields, software upgrades, inadvertent disabling or editing of rules and malfunctions of external systems commonly contribute to these malfunctions and current approaches for preventing and detecting such malfunctions are inadequate. In this presentation, I will describe results of an NLM-funded research project to catalog CDS malfunctions and identify strategies for preventing or mitigating them, including statistical methods for anomaly detection designed to quickly identify potential CDS malfunctions.
Bio: Adam Wright is an Associate Professor of Medicine at Harvard and a Senior Scientist at the Brigham and Women's Hospital. Dr. Wright's research focuses on making EHRs safer and more effective through better design, improved clinical decision support and harnessing clinical data to drive a learning healthcare system. In addition to research, Dr. Wright teaches Harvard's introductory biomedical informatics courses, and also teaches first medical students. He has a PhD in biomedical informatics from Oregon Health and Science University and a BS in mathematical and computational sciences from Stanford.
April 15, 2016 - Thomas Payne, MD. Associate Professor, Department of Medicine, University of Washington.
EHR 2020: Charting the Course to a Better Future
Asbstract: The Report of the AMIA EHR 2020 Task Force on the Status and Future Direction of EHRs articulates 10 recommendations to advance the value of electronic health record systems. The subject of a Senate Congressional hearing during the summer of 2015, this report provides concrete recommendations for how policymakers, providers and technology developers must work together to improve how we manage and use EHRs.
Bio: Dr. Payne is Medical Director of UW Medicine IT Services (since 2000), Associate Professor of Medicine, and Adjunct Associate Professor in Health Services and Biomedical Informatics & Medical Education. He is Attending Physician in Medicine at the University of Washington Medical Center and Harborview Medical Center. He serves as Board Chair of the American Medical Informatics Association. Dr. Payne attended Stanford University, the University of Washington School of Medicine, completed his internal medicine residency at the University of Colorado, and completed a fellowship at Massachusetts General Hospital in the Harvard Medical Informatics Fellowship program. He is board certified in Internal Medicine and Clinical Informatics.
April 8, 2016 - Nigam Shah, Ph.D. Associate Professor, School of medicine, Stanford.
Building a [Machine] Learning Healthcare System
Abstract: In the era of Electronic Health Records, it is possible to examine the outcomes of decisions made by doctors during clinical practice to identify patterns of care—generating evidence from the collective experience of patients. We will discuss methods that transform unstructured EHR data into a de-identified substrate for such evidence generation. We will review use-cases, which use the resulting de-identified data, to discover hidden trends, build predictive models, and drive comparative effectiveness studies in a learning health system.
Bio: Dr. Nigam Shah is associate professor of Medicine (Biomedical Informatics) at Stanford University, Assistant Director of the Center for Biomedical Informatics Research, and a core member of the Biomedical Informatics Graduate Program. Dr. Shah's research focuses on combining machine learning and prior knowledge in medical ontologies to enable use cases of the learning health system. Dr. Shah received the AMIA New Investigator Award for 2013, was elected into the American College of Medical Informatics (ACMI) in 2015 and is inducted into the American Society for Clinical Investigation (ASCI) in 2016. He holds an MBBS from Baroda Medical College, India, a PhD from Penn State University and completed postdoctoral training at Stanford University.
April 1, 2016 - Julia Adler-Milstein, Ph.D. Assistant Professor of Information, University of Michigan.
Interoperability: From the Chambers of Congress to the Front-lines of Care
have spent the past decade studying the progress of electronic health
information exchange (HIE) and EHR system interoperability in the U.S.
My talk will begin by setting the policy context and explaining how
federal policy efforts have sought to intervene to promote health
information exchange. Despite substantial resources devoted to HIE,
several recent evidence reviews reveal that we still lack a robust
understanding of the extent to which HIE improves care, and, more
importantly, the underlying mechanisms. To help address this gap, I will
present results from a recent study that assesses whether one approach
to HIE that has grown rapidly in recent years – Epic Systems’
CareEverywhere platform – is associated with improved care in the
Emergency Department, and whether this relationship is mediated by a key
hypothesized mechanism: the timeliness of information return.
Bio: Julia Adler-Milstein is an Assistant Professor at the School of Information with a joint appointment in the School of Public Health (Health Management and Policy). Her research focuses on policy and management issues related to the use of IT in healthcare delivery. Her expertise is in health information exchange and interoperability, with a specific focus on financial incentives and business models. She also studies the quality, productivity and efficiency of electronic health records. Julia graduated with a PhD in Health Policy from Harvard University. Prior to graduate school, she worked at the Center for IT Leadership at Partners Healthcare in Boston and in the Health and Life Sciences Division of Accenture.
March 11, 2016 - Tim Imler, MD - Assistant Professor of Medicine, Indiana University.
Tracking the Untrackable: Utilizing Natural Language Processing for Quality Monitoring from Clinical Documents
Abstract: Increasing utilization of electronic health records has led to an ever increasing availability for textual documents in electronic format. Procedural specialties are being assessed for quality due to the high cost and potential for poor outcomes. There has been a demand to track individual providers, however, much of the information is locked in “free text” and not accessible for large scale comparison. Recent research and development has shown the possibility for utilizing natural language processing for this tracking. In this talk, Dr. Imler will describe the background and how a system for procedural quality tracking is being created at the Regenstrief Institute utilizing NLP.
Bio: Dr. Timothy Imler is a board certified physician in Internal Medicine, Gastroenterology, and Clinical Informatics and works as an Assistant Professor of Medicine at Indiana University and is the Director of the Data Core at the Regenstrief Institute. His research focus has been on the utilization of natural language processing (NLP) to track individual providers of endoscopic procedures for quality and holds a patent for the algorithm for identifying quality in colonoscopy.
March 4, 2016 - Albert Chan, MD - Vice President, Chief of Digital Patient Experience at Sutter Health.
Leveraging informatics to transform the patient experience at Sutter Health
Abstract: Sutter Health is an integrated, not-for-profit healthcare delivery system serving over 100 geographically and economically diverse communities in Northern California. This talk will describe Sutter Health's efforts to leverage core health information technology solutions and evolving partnerships with early stage startups to enhance the patient experience.
Bio: Dr. Albert Chan, M.D., M.S. serves as Vice President, Chief of Digital Patient Experience for Sutter Health, where he is the responsible for developing a multi-phasic healthcare delivery strategy that engages and delights our diverse patient population and their families, and support operational and clinical leaders with the necessary tools, know-how, standards, and resources to optimize our patients’ digital experience. He is the corporate champion for My Health Online, which over 65% of Sutter Health’s patients leverage as a patient portal to view their health information, schedule appointments, and communicate with their doctors and other members of the care team. Albert also provides clinical leadership for the rapid development, testing and scaling of conceptual prototypes and Sutter Health’s proposed/signed technology investments. He currently co-chairs the Sutter Health Telemedicine Steering Committee. Previously, Dr. Chan was Chief Medical Information Officer (CMIO) of the Palo Alto Medical Foundation where he led a clinical informatics team that focused on the optimization of use of electronic health records (EHR) and personal health records (EHR) for patient care delivery and clinical operations. Additionally, Dr. Chan served as Co-Chair of California Health and Human Services Agency’s Patient Engagement Work Group, identifying innovative approaches to engaging and empowering patients and their families through the use of technology that harnesses the HIE infrastructure, and recommend how to incorporate these approaches into the State’s HIE services. He was also Co-Chair of the Certification Commission for Health Information Technology (CCHIT) Personal Health Records Work Group, which developed criteria for personal health record software certification. Dr. Chan earned a bachelor’s degree with honors in Biological Sciences from Stanford University and a medical degree from the University of California, San Diego School of Medicine. He completed his residency and year as chief resident in family medicine at the San Jose Medical Center Family Practice Residency, affiliated with Stanford University of School of Medicine. Albert concurrently completed a post-doctoral fellowship in Biomedical Informatics at Stanford University School of Medicine and Family Medicine Research at the University of California, San Francisco School of Medicine, studying the effects of point of care decision support in the management of hypertension. During fellowship, Dr. Chan also assisted with the launch of PAMFOnline, one of the earliest linked PHRs in the United States. Dr. Chan received the “Future Leader” Health Care Hero Award from the Silicon Valley/San Jose Business Journal in May 2012 and was a member of the 13th Cohort of the California Health Care Foundation Leadership Program, a two-year fellowship that offers clinically trained health care professionals the experiences, competencies, and skills necessary for effective vision and leadership of our health care system in the State of California. He was awarded the 2014 Epic PACAcademy Award, awarded annually to a physician member of Epic community, selected by his/her peers, in recognition of outstanding contributions to the Epic community.
February 26, 2016 - Phil Payne, Ph.D., - Professor and Chair, Department of Biomedical Informatics, Ohio State University.
Applied Biomedical Informatics: Data-Driven Approaches to Discovery and Practice
Abstract: Recent advances in the adoption and use of health information technology (HIT) in combination with increasing volumes of readily accessible and diverse biomedical data have had a growing impact on healthcare research and delivery. In many cases, this has led to new efficiencies, improved clinical outcomes, and enhanced levels of patient safety. In others, the impact has been less positive, and is associated with workflow and user experience dissatisfaction as well as perceptions of missed opportunities relative to the use of computational tools to enable data-‐driven discovery science and clinical decision-‐making. Given the confluence of these trends, it can be argued that now is the time to consider new ways in which the application of Biomedical Informatics theories and methods can advance health and wellness through the facilitation of data-‐driven paradigms for evidence generation and delivery. Achieving this type of paradigm is at the core of what has been recently promoted as “precision medicine”. Ultimately, the goal of such applied Biomedical Informatics solutions in the context of a “precision medicine” model can and should be to substantially improve human health and wellness through the optimal and efficient use of HIT tools and biomedical data by researchers, providers, patients, and community members. This seminar will explore ways in which the pursuit of such a goal requires us to re-‐think how we study HIT tools and data science in the broad biomedical domain in order to arrive at such data-‐driven approaches to discovery and practice.
Bio: Dr. Payne is Professor and Chair of the Department of Biomedical Informatics at The Ohio State University (OSU). He is also the inaugural Director of OSU’s Translational Data Analytics Initiative (TDA@OSU), a campus-‐wide program to create a singular presence in data analytics at The Ohio State University. He holds additional positions as an Adjunct Professor of Health Services Management and Policy within the OSU College of Public Health, Associate Director for Data Science within the OSU Center for Clinical and Translational Science (CCTS), and Co-‐Director of the Bioinformatics Shared Resources within the OSU Comprehensive Cancer Center (CCC). Dr. Payne is an internationally recognized leader in the field of clinical research informatics (CRI) and translational bioinformatics (TBI), and leads the Department of Biomedical Informatics Laboratory for Knowledge Based Applications and Systems Engineering (KBASE). His research portfolio is actively supported by a combination of NCATS, NLM, and NCI grants and contracts, as well a variety of awards from both non-‐ profit and philanthropic organizations. Dr. Payne received his Ph.D. with distinction in Biomedical Informatics from Columbia University, where his research focused on the use of knowledge engineering and human-‐computer interaction design principles in order to improve the efficiency of multi-‐site clinical and translational research programs. Dr. Payne’s leadership in clinical research informatics community has been recognized through his appointment to numerous national steering and advisory committees as part of the American Medical Informatics Association (AMIA), Association for Computing Machinery (ACM), National Cancer Institute (NCI), National Library of Medicine (NLM), and the CTSA consortium, as well as his engagement as a consultant to academic health centers throughout the United States and the Institute of Medicine. Dr. Payne is the author of over 175 publications focusing on the intersection of biomedical informatics and the clinical and translational science domains, including several seminal reports that have served to define a new sub-domain of biomedical informatics theory and practice specifically focusing upon clinical research applications.
February 19, 2016 - Qi Long, Ph.D., - Associate Professor, Department of Biomedical Informatics, Emory University.
Integrative Analysis of Transcriptomic and Metabolomic Data via Sparse Canonical Correlation Analysis with Incorporation of Biological Information
Abstract: It is of increasing importance to integrate different types of omics data to examine biological mechanisms in disease etiology. Canonical Correlation Analysis (CCA) provides an attractive tool to investigate such mechanisms. Traditional CCA methods use all available variables and, more recently, regularized CCA methods have been proposed to induce sparsity in the CCA vectors in order to yield interpretable results. It is well-known that variables in omics data are functionally structured in networks or pathways. We propose new sparse CCA methods that incorporate biological/structural information via undirected graphical networks. Our work is motivated by an in-vitro mouse toxicology study on the neurotoxicity of the combination of the herbicide paraquat and fungicide maneb in relation to Parkinson's disease (PD). We are interested in assessing association between transcriptomic and metabolomic data that may shed light on the etiology of PD. Our proposed methods use prior biological/structural information among genes and among metabolites to guide selection of important genes and metabolites in sparse CCA. Our simulations demonstrate that our structured sparse CCA approach outperforms several existing sparse CCA methods in selecting important genes and metabolites when structural information is informative and it is also robust to mis-specified structural information. Our analysis of the PD toxicology data reveals that a number of gene and metabolic pathways including some known to be associated with PD are enriched in the subset of genes and metabolites selected by our proposed approach. This is a joint work with Sandra Safo and Shuzhao Li.
Bio: Dr. Long is an associate professor and Director of Research in the Department of Biostatistics and Bioinformatics with a secondary appointment in the Department of Biomedical Informatics at Emory University. He earned his PhD in Biostatistics from University of Michigan in 2005 and is an elected member of International Statistical Institute. The thrust of his research is to advance statistical methodology and data analytics in medicine and public health with keen interests in precision medicine and implementation science and in big data. His interests in methodology research include big data methods (with applications to omics data and electronic health records data); missing data; causal inference; predictive modeling; Bayesian methods; functional data analysis; and clinical trials.
February 12, 2016 - Cinnamon Bloss, Ph.D., - Assistant Professor, Department of Psychiatry, University of California, San Diego, La Jolla, CA.
Impacts of Precision Medicine Advances on Individuals, Providers and Society
Abstract: The successful application of precision medicine advances to improve human health, in part, depends on the acceptance and use of such technologies by those they aim to help. This presentation will highlight empirical studies designed to evaluate the impacts of precision medicine advances - including genomics, microbiome testing, wearable sensors, and health big data - on individuals, families and healthcare providers. The studies presented utilize a variety of both quantitative and qualitative methods. Taken together, this work has shown a lack of adverse impact of precision medicine technologies, such as personal genomic testing, on individuals, which has been a concern raised by U.S. Food and Drug Administration officials. Results also suggest, however, a number of potential barriers to implementation and adoption of precision medicine advances in clinical and research settings. Some of these include difficulties related to patient–provider communication and a lack of genomics knowledge among physicians, the potential for genomics to exacerbate existing health disparities, the extent to which IRBs and other regulatory and policy frameworks are challenged by emerging biomedical advances, and the need to redefine concepts of privacy in an era of big data.
Bio: Dr. Bloss is an Assistant Professor at UCSD and an adjunct policy analyst at the J. Craig Venter Institute. Her research program focuses on the impacts of emerging biomedical technologies. Major projects have included empirical studies of the impact of direct-to-consumer genomics, clinical genome sequencing, biosensing and mobile health, and personal health big data. Dr. Bloss was recently awarded an R01 grant from the National Human Genome Research Institute’s (NHGRI) Ethical, Legal and Social Issues program to develop tools to understand privacy preferences of individuals who are exposed to health big data technologies. Dr. Bloss has published over 60 papers and mentored over 20 students, ranging in level from high school to post-doctoral fellow. Dr. Bloss is also a California-licensed clinical psychologist and has worked with adults, families and children with a wide range of issues.
February 5, 2016 - Ali Shojaie, Ph.D., - Assistant Professor, Department of Biostatistics, University of Washington.
Statistical Methods for Differential Network Analysis
Abstract: Recent evidence suggests that changes in biological networks, e.g., rewiring or disruption of key interactions, may be associated with development of complex diseases. These findings have motivated new research initiatives in computational and experimental biology that aim to obtain condition-specific estimates of biological networks, e.g. for normal and tumor samples, and identify differential patterns of connectivity in such networks, known as differential network analysis. In this talk, we will discuss new computational tools, based on advances in statistical machine learning, to jointly learn networks of interactions among diverse sets of biological measurements—e.g., genes, transcripts and proteins—in multiple biological conditions and to formally test whether the observed differences in interaction patterns are statistically significant or are due to randomness in estimation procedures.
Bio: Ali Shojaie is an Assistant Professor of Biostatistics at the University of Washington. Initially trained as an Industrial and Systems Engineer, Dr. Shojaie obtained his PhD in Statistics from the University of Michigan, while completing Masters degrees in Applied Mathematics and Human Genetics. Dr. Shojaie’s research lies in the intersection of machine learning for high-dimensions data, statistical network analysis and applications in biology and social sciences. He has developed methods for network-based analysis of various types of “omics” data, as well as methods for formal inference in high dimensions. Dr. Shojaie teaches multiple (regular and short) courses on machine learning and network analysis and is the Co-Director of the Summer Institute for Statistics of Big Data.
January 29, 2016 - Wenrui Dai, Ph.D.,- Postdoc, Department of Biomedical Informatics, University of California, San Diego, La Jolla CA.
Homomorphic Encryption Based Sampling and Statistic Computation for Genome-Wide Association Study
Abstract: Genome-wide association studies (GWAS) have been widely adopted in discovering the association between genotypes and phenotypes. Public cloud services provide a flexible way to mitigate the storage and computation burden in conducting GWAS, but unprotected disclosure of highly sensitive human genome data would lead to serious negative implications on individuals. Thus, data encryption has become attractive to guarantee privacy protection in collection, analysis and dissemination of human gnome data, which makes it crucial to develop efficient analytical algorithms on encrypted data. In this talk, we present secure protocols based on homomorphic encryption (HME) for sampling and statistic computation in GWAS. Homomorphic encryption is a kind of encryption technique that allows computation to be carried out directly on ciphertexts with no loss of accuracy. For rare variants study, HEALER framework is proposed to facilitate secure rare variants analysis with a small sample size. Secure rejection sampling and integer comparison are developed for homomorphic exact logistic regression model. We also presented a novel HME-based framework, FORESEE, to fully outsource statistic computation by enabling secure divisions over encrypted data. Two division protocols are introduced with a trade-off between complexity and accuracy in statistic computation. Experimental evaluations demonstrated the efficiency and effectiveness of the proposed protocols under the parallel implementation.
Bio: Wenrui Dai received B.S., M.S., and Ph.D. degree in Electronic Engineering from Shanghai Jiao Tong University, Shanghai, China in 2005, 2008, and 2014. He is currently a postdoctoral scholar with the Department of Biomedical Informatics, University of California, San Diego. His research interests include predictive modeling and learning-based image/video coding, image/signal processing.
January 22, 2016 - Chun-nan Hsu, Ph.D., - Associate Professor, Department of Biomedical Informatics, University of California, San Diego, La Jolla CA.
Weakly supervised learning of biomedical information extraction from curated data
Abstract: Numerous publicly available biomedical databases derive data by curating from literatures. The curated data can be useful as training examples for information extraction, but curated data usually lack the exact mentions and their locations in the text required for supervised machine learning. This paper describes a general approach to information extraction using curated data as training examples. The idea is to formulate the problem as cost-sensitive learning from noisy labels, where the cost is estimated by a committee of weak classifiers that consider both curated data and the text. We test the idea on two information extraction tasks of Genome-Wide Association Studies (GWAS). The first task is to extract target phenotypes (diseases or traits) of a study and the second is to extract ethnicity backgrounds of study subjects for different stages (initial or replication). Experimental results show that our approach can achieve 87 % of Precision-at-2 (P@2) for disease/trait extraction, and 0.83 of F1-Score for stage-ethnicity extraction, both outperforming their cost-insensitive baseline counterparts. The results show that curated biomedical databases can potentially be reused as training examples to train information extractors without expert annotation or refinement, opening an unprecedented opportunity of using “big data” in biomedical text mining.
Bio: Prof. Chun-Nan Hsu earned his Ph.D. in computer science from the University of Southern California, Los Angeles, CA, USA. He was Assistant Professor in computer science and engineering, Arizona State University, Tempe, AZ, USA before he joined Institute of Information Science, Academia Sinica, Taiwan, where he served as the project leader of the informatics group in the Advanced Bioinformatics Core, a core facility supported by the National Research Program of Genomics Medicine funded by the National Science Council, Taiwan, from 2005 to 2011. He joined the Information Sciences Institute, University of Southern California from 2009 to 2013 and moved to his current position at the Division of Biomedical Informatics, Department of Medicine, University of California, San Diego, La Jolla, CA, in November 2013. He has published nearly 100 peer-reviewed research articles in the fields of machine learning, data mining, and biomedical informatics. His team developed widely used software tools for biomedical sciences, including FASTSNP for functional analysis of gene variations, AIIAGMT for biological text mining, and various tools for cell image analysis. Prof. Hsu is a Senior Member of the Association of Computing Machinery and a professional member of the International Society of Computational Biology. He was elected as the President of the Taiwanese Association for Artificial Intelligence from 2009 to 2011. He won the IBM Faculty Award for his distinguished contributions to biomedical text mining in 2012.
January 15, 2016 - Rajiv Kumar, MD - Clinical Assistant Professor, School of Medicine, Stanford.
Utilization of Consumer Technology to Integrate Patient Generated Health Data in the EHR
Abstract: The electronic health record (EHR) is the home of patient variables, healthcare provider workflow, data/triage analytics, and the method to integrate these data and outcomes across institutional boundaries. Integration of patient generated health data in the EHR is vital on our path to effective individualized care plans and population health. Here we will discuss a diabetes management platform that leverage’s Epic’s patient portal app integration with Apple’s HealthKit platform on a patient’s mobile device, and the background workflow that requires no significant increase in time or effort by patients or healthcare providers alike.
Bio: Rajiv Kumar, MD is an Assistant Professor of Pediatric Endocrinology and Diabetes at the Stanford School of Medicine and Medical Director of Clinical Informatics at Stanford Children’s Health. With a focus on mobile health, his group works to improve passive communication and meaningful assessment of patient generated health data for chronic disease using sensors linked to mobile devices. Dr. Kumar’s group is equally interested in provider workflow, billing and policy related to this expanding model.
January 8, 2016 - Luca Bonomi, PhD - Posdoc, Department of Biomedical Informatics, University of California, San Diego, La Jolla, CA.
An Information-Theoretic Approach to Individual Sequential Data Sanitization
Abstract: Fine-grained, personal data has been largely, continuously generated nowadays, such as location check-ins, web histories, physical activities monitoring, etc. Those data sequences are typically shared with untrusted parties for data analysis and promotional services. However, the individually-generated sequential data contains behavior patterns and may disclose sensitive information. For example, the following two events detected in a hospital, i.e., a doctor leaves a patient's room and the doctor immediately enters a psychiatrist's office, might indicate that the patient is experiencing psychiatric problems. While a variety of sanitization techniques have been developed to prevent such privacy disclosure they often yield to released sequences with poor utility. In this talk, we present the problem of individual sequence data sanitization with minimum utility loss, given user-specified sensitive patterns. We propose a privacy notion based on information theory and sanitize sequence data via generalization. We show the optimization problem is hard and develop two efficient heuristic solutions. Extensive experimental evaluations are conducted on real-world datasets and the results demonstrate the efficiency and effectiveness of our solutions.
Bio: Luca Bonomi received B.S. and M.S. in Computer Engineering, University of Padova, and Ph.D. degree in computer science, Emory University in 2006, 2008, and 2015, respectively. He is currently a postdoctoral scholar with the Department of Biomedical Informatics, University of California, San Diego. His research interests include Data Privacy & Security, Sequential Pattern Mining, and Data Integration. His research work has been well received in prestigious Database, Data Mining, and Knowledge Discovery conferences.