Annotation is expensive but essential for clinical note review and clinical natural language processing (cNLP). However, whether computer-generated pre-annotation is beneficial to human annotation is still an open question. Our study introduces CLEAN (CLinical note rEview and ANnotation), a pre-annotation-based cNLP annotation system to improve clinical note annotation of data elements, and comprehensively compares CLEAN with a widely-used annotation system Brat Rapid Annotation Tool (BRAT). CLEAN includes an ensemble pipeline (CLEAN-EP) with a newly developed annotation interface (CLEAN-AT). A domain expert and a novice user/annotator participated in a comparative usability test by tagging 87 data elements related to Congestive Heart Failure (CHF) and Kawasaki Disease (KD) cohorts in 84 public notes. CLEAN achieved higher note-level F1-score (0.896) over BRAT (0.820) denoting significant difference in correctness (P-value < 0.001), with the mostly related factor as system/software (P-value < 0.001). No significant difference (P-value 0.188) in annotation time was observed between CLEAN (7.262 minutes/note) and BRAT (8.286 minutes/note), mostly related to note length (P-value < 0.001) and system/software (P-value 0.013). The expert reported CLEAN to be useful and satisfactory, while the novice reported slight improvements. CLEAN improves the correctness of annotation and increases usefulness/satisfaction with the same level of efficiency. Limitations include untested impact of pre-annotation correctness rate, small sample size, small user size, and restrictedly validated gold standard. CLEAN with pre-annotation can be beneficial for an expert to deal with complex annotation tasks involving numerous and diverse target data elements.
Chun-Nan Hsu, PhD, Associate Professor at the department of Biomedical Informatics, University of California, San Diego. Dr. Hsu has published more than 100 highly cited 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, leading to commercialized products. He was awarded Senior Member of Association of Computing Machinery (ACM) in 2011 and the IBM Faculty Award for his distinguished contributions to biomedical text mining in 2012.
November 03, 2017 - Md Momin Al Aziz, PhD student Department of Computer Science, University of Manitoba.
Privacy Preserving Search in Genomic Data Abstract:
Genomic data hold salient information about the characteristics of a living organism. However, there is a growing privacy concern regarding the collection, storage, and analysis of such sensitive human data. In this talk, I will talk about some specific issues and the potential solutions utilizing different cryptographic techniques.Bio:
Md Momin Al Aziz is a PhD student from Computer Science, University of Manitoba primarily working on different privacy and security issues of genomic data. He completed his Bachelors in Bangladesh University of Engineering and Technology and finished his MSc at CS, UofManitoba under the supervision of Dr Noman Mohammed.
October 27, 2017 - Tsung-Ting Kuo, PhD, Postdoc Fellow Department of Biomedical Informatics, UC San Diego.
The Blockchain Technologies: Introduction, Applications, and Platforms Abstract:
In this talk, I will introduce Blockchain technologies including their benefits, pitfalls, and the latest applications in the biomedical and healthcare domains. Also, I will discuss our most recent work to systematically review a set of Blockchain platforms, identify their technical features, and investigate the use cases of various technologies.Bio:
Dr. Tsung-Ting Kuo is a postdoc in UCSD DBMI. He earned his Ph.D. in CS from National Taiwan University. He was major contributor of UCSD DBMI team to win the ONC healthcare blockchain challenge, and NTU team to win four times of ACM KDDCup. His research focuses on Blockchain technologies.
October 20, 2017 - Christine Nibbelink, PhD, Postdoc Fellow Department of Biomedical Informatics UC San Diego.
Nurse Decision-Making and Clinical Decision Support in Acute Care Abstract:
The Institute of Medicine indicates that up to 98,000 deaths are linked with poor decision-making in the United States each year. Nurses make a decision every 30 seconds in acute care environments. Best patient outcomes require patient specific decision-making based on evidence. In addition, urgent changes in patient condition require time effective interventions for improved patient survival. Many factors influence nurse decision-making. Nursing clinical decision support systems are designed to facilitate use of evidence in combination with patient information to guide nurse decision-making in acute care environments. However, many nurses do not describe using evidence based clinical decision support in acute care nursing practice.Bio:
Christine Nibbelink finished her PhD in Nursing Informatics at University of Arizona. Currently, a postdoctoral fellow in the Department of Biomedical Informatics at UCSD, she is interested in decision-making, clinical decision support systems, and interprofessional communication in acute care environments. Postdoctoral research will include exploring interprofessional use of communication channels in acute care.
October 13, 2017 - Rebecca Marmor, MD, MAS, General Surgery Resident, UC San Diego.
Surgical Questions: Biomedical Informatics Answers
In this talk I describe two distinct questions posed by surgeons and the answers that biomedical informatics data sets and analyses provide. Surgeons have been puzzled by the recent trend towards bilateral mastectomy among breast cancer patients. I describe how we are able to utilize data from online health communities to understand the impact that online health communities may have on this trend, and to better understand the motivation of patients selecting this operation. Next, I describe the process by which surgeons were able to utilize readily available electronic health record data to assess compliance with post-splenectomy vaccination recommendations, and develop interventions to improve compliance.Bio:
Rebecca Marmor, MD, MAS is a general surgery resident at the University of California San Diego. She recently completed a two year NLM postdoctoral Biomedical Informatics Research fellowship. Her research interests include improving quality and safety for surgical patients and using data from online health communities to understand surgical decision-making of patients. She hopes to pursue a vascular surgery fellowship upon completion of residency.
October 06, 2017 - April Moreno, PhD, NLM Postdoc Fellow Department of Biomedical Informatics, UC San Diego.MS Patient Research Networks for Improving Diverse Health Outcomes Abstract:
Currently, over 80,000 members are participants of online multiple sclerosis health communities such as Health Care Journey, Accelerated Cure Network, the iConquer MS (MSpatient powered research network), and the Alliance for Research in Hispanic Multiple Sclerosis (ARHMS). MS patient research network (MSPRN) participation can be classified at different categories such as: Categories 1 (email list), Category 2 (patient-doctor community), Category 3 (MS-PPRN), and Category 4 (data sharing and blood samples). Of particular concern is the challenge of engaging a more numerous and diverse demographic range in Category 3 and 4 MS Patient Research Networks (MSPRN), which involve patient data sharing for research. In these two categories, participants share their data to improve minority MS health research, such as for researching effects of new therapies to delay disease progression. A culturally responsive approach is suitable for identifying predictive factors affecting diverse and inclusive participation and nonparticipation in Category 3 and 4 MSPRN. Although MS registries currently exist worldwide to collect MS data at the phenotype and genotype levels, there is still a lack of participation with diverse populations.Bio:
April Moreno completed her PhD from Claremont Colleges in Health Promotion Sciences and Information Systems and Technology. She is currently an NLM Postdoc Fellow at DBMI. She is interested in health IT policy, health disparities, and human computer interaction. She is currently working on multiple sclerosis patient research, and on health IT data privacy issues.
September 29, 2017 - Chunhua Weng, PhD, FACMI, Associate professor Department of Biomedical Informatics, Columbia University.
Enable Clinical and Translational Sciences with Clinical Research Informatics Abstract:
In this talk, Dr. Weng will give an overview of the field of clinical research informatics. On this basis, Dr. Weng will describe her ongoing research on combining public data in the ClinicalTrials.gov with electronic health records to facilitate clinical research knowledge acquisition and reuse and to optimize participant selection and accrual for clinical studies.Bio:
Dr. Chunhua Weng is an Associate Professor of Biomedical Informatics at Columbia University, where she has been a faculty member since 2007. She also co-leads the Biomedical Informatics Resource for the Columbia CTSA. Dr. Weng holds a Ph.D. in Biomedical and Health Informatics from University of Washington at Seattle. Dr. Weng has published > 120 papers to date with an H-index of 25. In recognition to her significant contributions to the field of biomedical and health informatics, Dr. Weng was elected as a fellow of the American College of Medical Informatics (ACMI) in 2015.
June 9, 2017 - Bo Luo, Ph.D., Associate Professor, Department of Electrical Engineering and Computer Science, University of Kansas.Healthcare Information and Social Networks: Data Utilization vs. Privacy
Abstract: Collecting patient information in clinical and pharmaceutical research is extremely expensive. On the other hand, with the rapid growth of Web 2.0 and social networks, we have observed a large number of patients sharing personal experiences in various healthcare-related social networks. We first present a Healthcare Forum Mining Ontology which defines health and patient related terms that can be extracted from the messages posted in online healthcare forums, along with the relationships between those terms. On the other hand, with the advancement of web technology, it becomes easier for malicious adversaries to pose real privacy threats. We argue that contents such as user attributes and unstructured text messages are critical in privacy protection. We will further discuss the risks associated with attributes and contents, in particular, attribute-reidentification attacks; and the potential of using social circles for privacy protection and automatic social circle detection.
Bio: Bo Luo is currently an associate professor with EECS department at the University of Kansas. He is the director of the Information Assurance Laboratory (IAL) at KU's Information and Telecommunication Technology Center (ITTC). He received Ph.D. degree from The Pennsylvania State University in 2008, M.Phil degree from the Chinese University of Hong Kong in 2003, and B.E. from University of Sciences and Technology of China in 2001. His recent works mostly lie in the intersection of data science and privacy and security. Dr. Luo has published 50+ refereed papers, including ones in top conferences and journals such as IEEE Oakland, ACM CCS, ACM Multimedia, IEEE TKDE, IEEE TIFS, IEEE TDSC, VLDBJ, etc.
June 2, 2017 -
Son Doan, Ph.D. and
Manabu Torii, Ph.D., Scientist Medical Informatics, Kaiser Permanente Southern California.
Overview of Current Clinical NLP: from Research Perspective to Practice
Abstract: In this talk we will present an overview of current clinical Natural Language Processing (NLP) under both views: academia and industry. We will first give an overview of several representative clinical NLP systems. Secondly, we will present basic components in an NLP system and discuss challenges in developing them in the clinical domain. Thirdly, we will present selected NLP tasks in more detail, including parsing, de-identification, and assertion classification. Finally, we will share some experiences in research and development of clinical NLP systems.
Son Doan, PhD, is a scientist in the medical informatics group at Kaiser Permanente Southern California. His area of interest is application of natural language processing in the biomedical and clinical domain. He received PhD in Computer Science from Japan Advanced Institute of Science and Technology in Japan and worked as a researcher at National Institute of Informatics in Japan and got a postdoc training at Vanderbilt University. Before joining the current group, he was a Programmer Analyst at Department of Biomedical Informatics, UC San Diego.
Bio: Manabu Torii, PhD, is a scientist in the medical informatics group at Kaiser Permanente Southern California. His area of interest is application of natural language processing in the biomedical and clinical domain. He received PhD in Computer Science from the University of Delaware. Before he joined the current group, he was a research assistant professor of Computer Science at the University of Delaware.
May 26, 2017 -
Yuan Luo, Ph.D., Assistant Professor, Division of Health and Biomedical Informatics, Northwestern University.
Using Machine Learning to Predict Laboratory Test Results
Abstract: Clinical decision support that integrates multiple elements of laboratory data could be highly useful in enhancing laboratory diagnosis. Using the analyte ferritin in a proof-of-concept, we extracted clinical laboratory data from patient testing and applied a variety of machine learning algorithms to predict ferritin test result using the results from other tests. We show that predicted ferritin results may sometimes better reflect underlying iron status than measured ferritin. We next integrate temporality into predicting multi-variate analytes. We devise an algorithm alternating between cross sectional imputation and auto regressive imputation. We show modest performance improvement of the combined algorithm compared to either component alone.
Bio: Yuan Luo is an Assistant Professor at Department of Preventive Medicine, Division of Health & Biomedical Informatics with courtesy appointments in IEMS and EECS, all at Northwestern University. He holds a PhD degree from MIT EECS. His research interests include machine learning, natural language processing, time series analysis, and computational genomics, with a focus on medical applications. He proposed Subgraph Augmented Non-negative Tensor Factorization (SANTF) for building a clinical model that improves both accuracy and interpretability, by turning EHR into graph representations and applying tensor factorization to mining graph features. He has extended SANTF to time series analysis and computational genomics.
May 19, 2017 - Gina Merchant, Ph.D., Postdoc Researcher, DBMI, UCSD.
Identifying Online Conversations about Health and How Information Travels in Social Networks: Anti-Vaccine Sentiment as a Health Threat
Abstract: Despite the success of vaccines nationally and around the world, vaccine hesitancy is on the rise. Outbreaks of diseases such as measles indicate a profound public health need to understand why individuals are vaccine hesitate, and how to increase confidence in vaccines. Parents’ reluctance to accept the childhood vaccine schedule is encouraged by two factors. First, today’s postmodern medical paradigm encourages shared decision making, which empowers the patient to have a voice in her care. Second, Web 2.0 culture supports multiple ways of “knowing,” and user-generated content advances personalized truths that may be weighted equally to scientific consensus. My K application uses childhood vaccination schedules as a case study, examining how exposure to non-credible health information may cluster and spread in online networks, and will test how patient-decision aids may inoculate attitudes about a controversial topic against real-world misinformation. This project represents the first step in a program of research that seeks to uncover how online connectivity and information exposure affects real-world healthcare utilization and health behavior.
Bio: I am a behavioral scientist whose research is at the intersection of psychology, public health informatics, and data science. I am interested in quantitatively and qualitatively measuring how spending time on social media platforms, creating and engaging with content, affects our health. I seek to uncover methods to conduct large-scale content analysis of unstructured text data that is exchanged in online networks. Once we are able to interpret these data, I believe we will have a better understanding of how to intervene in virtual spaces to promote health-enhancing behaviors, and better predict who is at risk for health impairing behaviors. I want to discover ways to leverage social media and other Web 2.0 technologies to improve individuals’ health, and better understand how on- and offline social networks synergistically/antagonistically influence health attitudes and behaviors. I completed my PhD in Public Health in 2015, and hold a masters degree in Experimental Psychology. I am currently an NLM-funded postdoctoral fellow in the Division of Biomedical Informatics here at UCSD.
May 12, 2017-
Yuanyuan Fang Ph.D., Student, University of Maryland, Baltimore Count.
Actions as a Source for Process Common Ground
Abstract: <no abstract for Yuanyuan’s talk>
Bio: Yuanyuan Feng is a visiting graduate student from University of Maryland, Baltimore County, where she is a member of the Bodies in Motion Lab working with her advisor Helena Mentis. She graduated with a M.B.B.S. at Tianjin Medical University in China in 2011, and a M.S. in Health Informatics at University of Minnesota, Twin Cities in 2013. Her interests lie in biomedical informatics, human-computer interaction (HCI) and computer-supported cooperative work (CSCW). Her current research focuses on the team knowledge sharing in the operating rooms.
Is FHIR a Viable Messaging Standard to Model Heart Failure Readmission? An Evaluation of Clinical, Community, and Telehealth Data
Abstract: Technological innovation within the healthcare system is a complex and challenging process. Despite early success within the laboratory setting, market and regulatory concerns must be considered to translate these results into meaningful clinical care. Identification and engagement of relevant stakeholders ranging from insurers to practicing clinicians and governmental regulators are necessary for product success. Using the lens of novel diagnostic tests, this talk will identify the challenges biomedical innovators face in the translation between the lab and the clinic.
Bio: Kevin is a first-year Biomedical Informatics student at UCSD, whose research emphasizes that lies at the intersection of implementation science and healthcare innovation. Prior to starting his graduate studies, he worked as a clinical cancer data analyst, where his work has contributed to the identification of tumor biomarkers for guiding clinical cancer care. Subsequently while at UCLA, Kevin received his MS in Health Policy and Management, where his studies focused on health economics and the commercialization of healthcare innovations. Currently, Kevin works with the UCSD Consulting Club aiding in the strategic alignment of biotechnology startups with the clinical and regulatory stakeholders.
May 5, 2017-
Robert El-Kareh M.D., Associate Clinical Professor, DBMI, UCSD.
Using Health IT to Improve Diagnostic Safety at UCSD
Abstract: Diagnostic errors represent a significant and under-recognized threat to patient safety. In 2015, the National Academy of Medicine published the report Improving Diagnosis in Health Care which described the scope of the problem and highlighted health information technology (HIT) as a crucial part of the solution. In this presentation, I will discuss a few of the current efforts to improve diagnostic safety at UC San Diego Health using HIT and outline areas with the potential to yield important results in the near future.
Bio: Robert El-Kareh, MD, MPH, MS is a clinical informatician in the Divisions of Biomedical Informatics and Hospital Medicine and Information Services at UCSD. He is a medical director of Quality Improvement and Clinical Decision Support and a practicing hospital medicine physician at UC San Diego Health. His primary research interests involve the use of electronic data to identify and prevent diagnosis errors in healthcare. Dr. El-Kareh's expertise is in the design and implementation of clinical decision support that is integrated within the workflow of front-line clinicians. In addition, he chairs the Clinical Decision Support Committee and leads several projects related to patient safety.
April 28, 2017-
Lian Jian Ph.D., Assistant Professor, University of Southern California.
Uncertainties in Crowdsourcing Contests of Creative Solutions
Abstract: Crowdsourcing contests are contests by which organizations outsource tasks to talents on the Internet. Due to anonymity and lack of trust, participants in such (typically winner-take-all) contests face uncertainties (e.g., contest holder not paying after receiving submissions). We examine how the use of third-party prize guarantees (guaranteeing that a winner will be paid) and in-process feedback (numeric ratings to individual designs and public textual comments) can help reduce such risks, thereby attracting submissions. We find that prize guarantees increase submissions but has no effect on the number of participants. Higher volume of in-process ratings lead to more submissions and more participants, and such an effect is bigger in contests without prize guarantees. High ratings to individual submissions discourages future submissions, and this negative effect is stronger in non-guaranteed contests.
Bio: Lian Jian is an assistant professor in the School of Communication at USC. She obtained her Ph.D. from the University of Michigan, where she studied the economics of information. Her research focuses on the economics of online information systems, using multiple research methods. Examples of her past research include a game-theoretic model of crowdsourcing contests, lab experiments on manipulation by bidders in prediction markets, an econometric study of choice of feedback provision strategy by users on eBay, and the diffusion of rumors during the 2012 US presidential election.
April 21, 2017-
Dennis Paul Wall M.D., Associate Professor, Department of Pediatrics, Stanford University.
Machine Learning Approaches to Disentangle the Complexity of Autism
Abstract: In this talk I will describe my work in machine learning to mobilize the ex-clinical detection of developmental delays in children to broaden reach and decrease the average age of diagnosis for earlier delivery of therapy. I will describe our complementary work on an artificial intelligence wearable designed to develop social skills in children with autism through computer vision technologies. Finally, I will describe how these digital phenotyping approaches can couple with large-scale genomic sequencing to enable a clearer picture of the forms of autism.
Bio: Dr. Dennis P. Wall, PhD is Associate Professor of Pediatrics, Psychiatry and Biomedical Data Science at Stanford Medical School. He leads a lab in Pediatric Innovation focused on developing methods in biomedical informatics to disentangle complex conditions including autism and related developmental delays. His work includes deep learning approaches to tie genotype to phenotype and mobilized AI to deliver therapy and measure progress in children with developmental delays. He has received numerous awards including the Harvard Medical School Leadership award and the Slifka/Ritvo Clinical Innovation Award for outstanding advancements autism research.
April 14, 2017-
Olivier Harismendy, Ph.D., Assistant Professor, DBMI, UCSD.
Accessing and Sharing Cancer Genomics Data
Abstract: Recent advances in technology have enabled the widespread generation of genomics and genetics data in cancer research and care. This large resource is however under-utilized for research and additional insights into cancer etiology, progression and response to treatment may be drawn by aggregating data from multiple sources, extracting additional molecular information or mapping it to more refined phenotypic information. I will present the current solutions and ongoing efforts pursued by the cancer genomics community to distribute and share these dataset. I will introduce the standards being developed to represent and exchange genomic data. Finally, I will discuss local efforts at the UCSD Moores Cancer Center to distribute and facilitate the use of the data generated as part of clinical care.
Bio: Dr. Harismendy is an Assistant Professor of Medicine at UC San Diego. He leads the Oncogenomics laboratory at Moores Cancer Center and his main interest is the development of genomics assays and computational analysis for personalized cancer care. In the past 15 years, Dr Harismendy’s research has contributed to the understanding of transcriptional regulation and the role of regulatory variation in human diseases. He has developed methods to characterize the molecular and cellular heterogeneity of tumors. He currently studies the contribution of epigenetic and genetic variations to cancer susceptibility and to the development of drug resistance.
April 7, 2017-
William Knox Carey, Ph.D., Vice President, Healthcare, Intertrust/General Manager, Genecloud.
Taking Patients Rights Seriously: Ten Challenges
Abstract: Progress in biomedical research is fueled by increasingly large data sets. It is important to recognize, however, that every data point in a study derives from a person who has certain rights in the data: the right to participate or not, the right to learn about the results of the study, the right to control how their data are used. In the past, we have tended to minimize or avoid these complex questions of patient rights. In this talk, I will discuss ten specific challenges -- technical and non-technical -- that confront us as we seek to turn patients into research partners.
Bio: Knox Carey is Vice President of Healthcare Initiatives at Intertrust Technologies Corporation. He is General Manager of the Genecloud project, which aims to balance individual privacy with access to sensitive genomic and other healthcare data. Knox is also a leading member of the Security Working Group at the Global Alliance for Genomics and Health, an organization defining standards and best practices for data sharing in genomics. Dr Carey is a three-time graduate of Cornell University (BS 92, MEng 94, PhD 99), where he majored in Electrical and Computer Engineering with concentration in digital signal processing, information theory, and applied mathematics.
Rui Chen, Ph.D., Research Scientist, Samsung Research America.
From Academia to Industry: A Preview of Data Science
Abstract: Data science has been a buzzword in both academia and industry. In this talk, we will try to bridge the gap between academia and industry in the context of data science. We will explore the roles of data science in industry and show how to apply school knowledge to solve a real-world business problem on user growth. We demonstrate how to combine various computer science techniques to acquire, engage and retain users under a business funnel model.
Bio: Rui is a senior staff research scientist at Samsung Research America, where he leads a horizontal data science team that supports multiple Samsung products, including Samsung Pay and Samsung Health. He has published nearly 40 papers in top venues on databases, data mining and data security, such as PVLDB, VLDBJ, TKDE, ICDE, SIGKDD, CCS, ICDM, SDM, and CIKM. His papers have been cited more than 2,000 times. Prior to his post at Samsung, he was a research assistant professor in the Department of Computer Science at Hong Kong Baptist University and a postdoctoral fellow at the University of British Columbia. He was the recipient of CIKM 2015 Best Paper Runner Up and the Alexander Graham Bell Canada Graduate Scholarship issued by Natural Sciences and Engineering Research Council of Canada.
March 10, 2017-
David Classen, M.D., CMIO, Pascal Metrics Inc.Harnessing Health IT to Improve Patient SafetyAbstract:
This presentation will focus on the safety of Health IT Systems. A recent IOM report focused on Health IT and Patient Safety questioned whether the broad adoption of Health IT in the last decade had actually improved the safety of care and cited instances of where these systems had actually harmed patients. This presentation will review work on assessing safety improvements with Health IT and what the future for Health IT and safety may look like.Bio:
David C. Classen, M.D., M.S. Dr, Classen is the CMIO at Pascal Metrics, a Patient Safety Organization (PSO) and an Associate Professor of Medicine at the University of Utah and an Active Consultant in Infectious Diseases at The University of Utah School of Medicine in Salt Lake City, Utah.
He received his medical degree from the University Of Virginia School Of Medicine and a Masters of Science degree in medical informatics from the University Of Utah School Of Medicine. He served as Chief Medical Resident at the University of Connecticut. He is board certified in Internal Medicine and Infectious Diseases. He developed the medication safety programs at Intermountain Healthcare; He was the chair of Intermountain Health Cares Clinical Quality Committee for Drug Use and Evaluation and was also the initial developer of patient safety research and patient safety programs at Intermountain Healthcare. In addition he developed, implemented and evaluated a computerized physician order entry program at LDS Hospital that significantly improved the safety of medication use.
He was a member of the Institute of Medicine Committee (IOM) that developed the National Healthcare Quality Report and he was also a member of the Institute of Medicine Committee on Patient Safety Data Standards. He was recently a member of the Institute of Medicine Committee on Health Information Technology and Patient Safety.
He chaired the QUIC (Federal Safety Taskforce)/IHI Collaborative on Improving Safety in High Hazard Areas. Dr. Classen was Co Chair of the Institute of Healthcare Improvements Collaborative on Perioperative Safety and the Surgical Safety Collaborative at the Institute of Healthcare Improvement (IHI). He was also a faculty member of the IHI/National Health Foundation Safer Patients Initiative in the United Kingdom. In addition Dr. Classen is one of the developers of the “Trigger Tool Methodology” at IHI, used for the improved detection of adverse events which is currently being used by more than 500 different healthcare organizations through out the Unites States and Europe.
He currently co chairs the National Quality Forum’s AHRQ Common Formats Committee and Dr. Classen is an advisor to the Leapfrog Group and has developed and implemented the CPOE/EHR flight simulator for AHRQ and National Quality Forum. This Electronic Health Record (EHR) Flight simulator has been used to evaluate hundreds of inpatient and ambulatory EHR systems after implementation across the United States and The United Kingdom and is a critical part of the National Quality Forum’s Safe Practice #16 for Computerized Provider Order Entry within EHRs.
David Page, Ph.D., Professor, Department of Biostatistics and Medical Informatics, University of Wisconsin-Madison
High-Throughput Machine Learning from EHR Data
Abstract: The widespread use of electronic health records and the many recent successes of machine learning raise at least two natural questions. How well can future health events of patients be predicted from EHR data, at various lengths of time in advance? And how can such predictions improve human health? This talk answers the first question via a new approach called "high-throughput machine learning," and it speculates about answers to the second question. In particular, this talk argues that many healthcare applications require not just accurate prediction, but accurate prediction by causally-faithful models. Causal discovery from observational data is already a major research direction in machine learning and statistics, and this talk discusses new approaches across the spectrum from when "we know all the relevant variables" to when "we know only one relevant variable" for the task at hand. If time permits, the talk will also touch on the issue of protecting patient privacy while empowering the construction of accurate predictive models.
Bio: David Page is a Vilas Distinguished Achievement Professor at the University of Wisconsin-Madison. His primary appointment is in the Dept. of Biostatistics and Medical Informatics in the School of Medicine and Public Health, with an appointment in the Dept. of Computer Sciences where he teaches machine learning. His PhD in CS is from the University of Illinois at Urbana- Champaign, and he became involved in biomedical applications of machine learning as a post-doc in what was then the Computing Laboratory at Oxford University. He directs the Cancer Informatics Shared Resource of the Carbone Cancer Center and is a member of the Genome Center of Wisconsin. He previously served on the NIH's BioData Management and Analysis Study Section and the scientific advisory boards for the Wisconsin Genomics Initiative and the Observational Medical Outcomes Partnership, as well as the editorial boards for Machine Learning and Data Mining and Knowledge Discovery. He currently is on the National Library of Medicine Study Section (BLIRC) and directs the EHR project within UW-Madison's BD2K Center for Predictive Computational Phenotyping.
Mary Devereaux, Ph.D., Assistant Director, Research Ethics Program, UCSD
Patient Consent in the Era of Big Data
Abstract: With the move to electronic health records (EHRs) and the growing capacity to gather and process terabytes of medical information, researchers understandably wish to access aggregated data to analyze patient outcomes, health demographics, and the economics of health care (Safran et al. 2007). But all medical information, whether sensitive or not, is governed by legal requirements for privacy, confidentiality, and security, e.g., HIPAA. The use of patient data for research also raises a host of ethical issues. Some patient information may be highly confidential, including family history, genetic testing results, a diagnosis of addiction, or immigration status. This talk examines ethical issues in the secondary use of personal health information gathered for clinical purposes. In particular, discussion will focus on what patients understand — and expect — regarding the de-identification and security of their information.
Bio: Mary Devereaux, Ph.D., is a philosopher and bioethicist at University of California, San Diego (UCSD). She is Assistant Director of the UCSD Research Ethics Program and the San Diego Research Ethics Consortium, and Director of the UCSD Biomedical Ethics Seminars. She also holds an appointment in the Health Law & Policy Program, where she is Academic Coordinator, and Adjunct Professor of Law at California Western School of Law. Devereaux serves on the Hospital Ethics Committee at the Medical Center Hillcrest and provides ethics training in the School of Medicine and a variety of graduate programs in health sciences. She is founder and Director of Tough Cases, a monthly case-based ethics discussion for clinicians and medical staff, and Co-Director of the Medical Humanities Research Group on campus. She speaks widely on issues in biomedical and research ethics for academic and lay audiences. Recent publications include work on ethical and regulatory issues in stem cell research, reproductive medicine, cosmetic surgery, and medical tourism. Devereaux is a member of the American Society for Bioethics and Humanities and the American Philosophical Association.
February 17,2017 - Michael Hogarth, M.D., Professor, Department of Pathology and Laboratory Medicine, UC Davis Medical Center EDRSThe OneSource Initiative: An Approach to Structured Sourcing of Key Data in Electronic Health Records
Some of the most widely used electronic health record (EHR) systems in the US have significant shortcomings as information systems. Current systems have three key deficiencies regarding sourcing and managing of key data. First, key data elements are difficult to find in the record, or not in the record at all. Second, when key data is in the record, it may have multiple different states/values in different places, making it difficult to determine which one is “true". Third, the majority of key data elements for the appropriate management of major conditions are in unstructured text, making it is difficult to use these data for computer-assisted interventions. In addition to these shortcomings for key data, "core data element collections" for particular conditions have not been standardized in order to ensure they exist in the record. The OneSource Initiative is a collaborative project involving UCSF, FDA, CDISC and ONC to address these shortcomings and render core key data elements in computable form to support a broad range of processes such as coordination of care, population health analytics, and clinical research.Bio:
Dr. Hogarth is a board certified in Internist in the UC Davis Division of General Medicine and health informatics faculty in the Department of Pathology and Laboratory Medicine. He has taught health informatics at UC Davis for over 20 years. He also currently serves the UC Davis Health System as informatics and data quality lead for the Healthcare Analytics group. Dr. Hogarth has been involved in several large-scale informatics initiatives including California Electronic Death Registration System, the Athena Breast Health Network project, the I-SPY2 adaptive breast cancer clinical trial, UC-ReX, pSCANNER, and the California Precision Medicine Consortium. In collaboration with UC Santa Barbara, his team is currently building a new Birth Registration system for California.
February 10, 2017-
Kai Zheng, Ph.D., Associate Professor, Department of Informatics, UCI
Electronic Health Records Adoption and the Implications for Health Services Research
Abstract: Most U.S. hospitals and clinics have by now implemented electronic health records (EHR) as a result of the recent policy mandate. While adoption of EHR introduces numerous benefits (e.g., elimination of illegible handwritten notes), it is also associated with a number of unintended adverse consequences, including a detrimental effect on the quality of clinical data recorded. Some notable causes for this effect are coexistence of paper forms and EHR; clinicians’ lack of knowledge of, or enthusiasm in, entering data in a rigid, structured format; and inaccurate coding due to a prominent emphasis on revenue management in many EHR systems currently in use. In this talk, I will showcase some manifestations of the diminished quality of clinical data in the post-EHR era, and their potential impact on clinical, translational, and health services research.
Bio: Kai Zheng PhD is Associate Professor of Informatics in the Department of Informatics at the University of California, Irvine (UCI). He co-directs the Center for Biomedical Informatics at the UCI Institute for Clinical and Translational Science. Zheng’s research draws upon techniques from the fields of information systems and human–computer interaction to study the use of information, communication, and decision technologies in patient care delivery and management. His recent work has focused on topics such as interaction design, workflow and sociotechnical integration, and diffusion and evaluation of health IT. Zheng received his PhD degree in Information Systems from Carnegie Mellon University.
February 3, 2017-
Jejo Koola, M.D., Assistant professor, DBMI, UCSD
Principles and Practices of Data Visualization in Medical Informatics
Abstract: The complexity of data has grown exponentially over the last 30 years. These data allow us to predict disease, identify treatments, and make prognoses. However, with the ever larger datasets also come ever more complicated models that stretch the abilities of human cognition. A central task in information visualization is to find the appropriate visualization paradigm for both the data and the problem scenario at hand. We will explore several central paradigms for the visualization and analysis of healthcare data. In addition, we will demonstrate practical toolsets to employ for these visualization tasks.
Bio: Jejo Koola is a practicing internist in the field of hospital medicine and clinical informatics. He received his medical degree from the Medical University of South Carolina and completed his residency at the Medical College of Virginia. He completed a fellowship in Biomedical Informatics through the Department of Veterans Affairs in conjunction with Vanderbilt University. His is focused on using informatics tools (including predictive analytics, natural language processing, and information visualization) to improve the care of multi-morbid hospital patients. He has published in several clinical and informatics journals.
Jing Zhang, NLM Ph.D. fellow, DBMI, UCSD
Supporting Information Needs of Transitional Phases in Diabetes Management in Online Health Communities
Abstract: As of 2014, 29.1 million people in the US have diabetes. Diabetes has a substantial and increasing impact on the quality of life. Patients face the burden of self-management and the challenge of ‘transitional’ phases, when they need to find out about their options and the next course of action. The field has under-explored the specific information needs patients have during those transitional phases. I aim to investigate the information needs in the transitional phases and develop design requirements in providing balanced and comprehensive information to better support patient information needs.
Bio: I am a PhD candidate at DBMI. My research interests lie in the intersection of health informatics and human-computer interaction. I am specifically interested in studying how clinicians, patients and health consumers use technology, with the goal of improving system design and human performance. I am currently working with Dr. Jina Huh on online health communities and will also start on a usability project.
Zhanglong Ji, Ph.D. fellow, DBMI, UCSD
Beacon Service Against Bustamante Attack
Abstract: Genome data sharing has been strictly confined due to privacy concerns. Recent researches have proven that even releasing existence of alleles in a database may leak the existence of a person. In this presentation, I will present three algorithms to protect data privacy given different assumptions. In all three algorithms, the attackers have genome information of a patient whom he wants to know whether is in the database. The first algorithm assumes attackers do not have any information on allele frequency; the second one assumes attackers know those frequencies, and have correct genome information. The last one assumes attackers have both allele frequencies and genome frequencies with some random errors.
Bio: Zhanglong Ji is a PhD student in Computer Science program right now. His research interest is in machine learning and privacy protection. Before coming to UC San Diego, he majored in Statistics in Peking University.
January 20, 2017-
Mike Zaroukian, M.D., Ph.D., Vice President & CMIO Sparrow Health SystemElectronic Health Record Usability and Healthcare's "Quadruple" AimAbstract:
Achieving the “Triple Aim” of improved patient experience, improved population health, and reducing per-capita health care spending in an era of healthcare delivery and payment transformation requires a synergy between people, processes and technologies to ensure that every patient receives high quality and high value care, every time. Over the past 6 years, the incentives for rapid adoption and expectations for robust use of electronic health record (EHR) systems and other health IT has been accompanied by increasing reports of physician burnout and complaints that poor EHR usability is contributing to the problem. This has prompted a call for a fourth aim related to improving clinician satisfaction. To facilitate achievement of this so-called “Quadruple Aim”, health IT professionals should be grounded in the concepts of EHR usability, user experience, and user-centered design from the clinician’s perspective. In this presentation, we will discuss health IT usability, why it matters, developing a framework to approach EHR usability challenges and opportunities, and how clinicians and IT professional can work together to improve usability in support of the “Quadruple Aim”.Bio:
Dr. Zaroukian is VP & CMIO at Sparrow and a Professor of Medicine at Michigan State University. A practicing primary care physician board-certified in internal medicine and clinical informatics, he provide physician executive leadership to EMR implementation and optimization, along with other health IT capabilities to transformation care. Dr. Zaroukian has published research and given hundreds of national and international presentations on the use of health IT to improve patient care, research, education and administration. He is a former residency director and previous recipient of the HIMSS Physician IT Leadership Award, a "Top 25 Clinical Informaticists" award, and an AMIA Leadership Award. He is current Chair of the HIMSS North America Board of Directors and Chair-elect of the global HIMSS Board of Directors. He is a former member of the American Medical Association (AMA) Health Information Technology Advisory Group, past chair of the American College of Physicians Medical Informatics Committee, and served on the ONC Health IT Policy Committee Advanced Health Models and Meaningful Use Workgroup. He currently serves on the “Optimizing Strategies for Clinical Decision Support” workgroup for the National Academy of Medicine Leadership Consortium for a Value & Science-Driven Health System.
January 13, 2017-
Nansu Zong, Ph.D., Postdoctoral Fellow, DBMI, UCSD
Predict Novel Drug-Target Associations Based on Heterogeneous Networks of Biomedical Linked Data
A heterogeneous network topology possessing abundant interactions between biomedical entities has yet to be utilized in similarity-based methods for predicting drug-target associations based on the array of varying features of drugs and their targets. We introduce a similarity-based drug-target prediction method that enhances existing association discovery methods by using a topology-based similarity measure. DeepWalk, a deep learning method, is adopted in this study to calculate the similarities within a tripartite heterogeneous network generated from the biomedical linked data. With the execution of the four-step experimental design, this proposed method shows promising results for drug-target association prediction.
Bio: Dr. Zong is a postdoctoral researcher in department of biomedical informatics, school of medicine, university of California, San Diego. He obtained the Ph.D. in computer science and engineering from Seoul National University. His research focuses on Big Data, Semantic Web and Linked Data, Data processing, management, and mining in Biomedical domain.