2021 Health Informatics Research Seminars

Duke

Broadcast Link: Seminar

Abstract:

The N3C project is designed to simplify access to EHR-derived data on individuals screened for COVID-19 at N3C member sites. Data from more than 60 sites using four different data models are submitted, harmonized, and transformed into the OMOP 5.3 data model and ingested into the secure N3C Data Enclave. This seminar will address the security and governance of N3C, the ingestion process, and the types of clinical scenarios and discoveries being investigated in N3C.

Biosketch:

Warren A. Kibbe, PhD, FACMI, is Vice Chair and Professor of Biostatistics and Bioinformatics at Duke University, the Informatics lead for the Duke Clinical and Translational Sciences Institute, and the Chief Data Officer for the Duke Cancer Institute. His research interests include data representation for clinical trials, especially improving the computability and interpretability of biomarker and eligibility criteria; data interoperability between medical records and decision support algorithms; improving data representation and interoperability for biomedical research using ontologies, developing novel analysis and visualization tools for next gen sequencing data, especially methylseq. Prior to joining Duke, he served as an acting deputy director of the NCI and was the director of the NCI’s Center for Biomedical Informatics and Information for four years.  He was one of the architects of the Genomic Data Commons initiative, which was the NCI’s foray into creating a highly accessible and highly accessed cancer data repository for clinical, proteomic, imaging and genomic data. Dr. Kibbe has been a proponent for open science and open data in biomedical research and helped define the data sharing policy for the NCI Cancer Moonshot program. He also helped architect the joint NCI-DOE computational and biomedical research collaboration.  Dr. Kibbe is the co-Founder of the Cancer Informatics for Cancer Centers (http://Ci4CC.org) society, and through Ci4CC organized twice-yearly meetings of cancer informatics faculty and leaders from the majority of NCI-designated Cancer Centers. Dr. Kibbe is one of the three MPIs for the NIH RADx-UP Collection and Data Coordination Center. The National Institutes of Health (NIH) Rapid Acceleration of Diagnostics for Underserved Populations (RADx-UP) aims to ensure that all Americans have access to COVID-19 testing, with a focus on communities most affected by the pandemic. Dr.Kibbe is also an MPI for the NHGRI-funded FUNCTION Center, one of the centers for excellence in genomic science.  Dr. Kibbe is currently a part of the NCATS National COVID Cohort Collaborative (N3C) where he is engaged in the Portals and Dashboards subgroup of the Collaborative Analytics Workstream.

UNC-CH

Broadcast Link: Seminar

Abstract:

Computable phenotypes are patient cohort definitions, translated into code; algorithms used to query EHR data to identify patients meeting certain criteria. Though computable phenotypes are common tools in the field of clinical informatics, they are notoriously difficult to share across multiple institutions, and may suffer from semantic ambiguity—what Researcher A defines as the “right” set of diagnosis codes for a condition of interest may differ wildly from Researcher B. This seminar explores whether clinical ontologies (like SNOMED CT and the Human Phenotype Ontology), and semantic web technologies and standards can address these challenges of interoperability and ambiguity, and serve as a step towards more sharable, reproducible computable phenotypes.

Biosketch:

Dr. Emily Pfaff is Administrative Director for Informatics and Data Science at NC TraCS Institute, UNC Chapel Hill’s Clinical and Translational Science Award. She has a PhD in Health Informatics and a master’s in Information Science, both from UNC Chapel Hill. Her primary expertise and research interests are in computable phenotyping and clinical data modeling in support of translational research.

UNC-Charlotte

Broadcast Link: Seminar

Abstract:

This seminar will focus on the aim to combine low-dimensional clinical and lab testing, as well as high-dimensional CT imaging data to accurately differentiate healthy individuals, COVID-19 and non-COVID viral pneumonia patients, especially at early stage of infection. For this study, 214 non-severe (NS) and 148 severe (S) COVID-19 patients, 198 non-infected healthy (H) participants and 129 non-COVID viral pneumonia (V) patients were recruited. The participants’ clinical information (23 features), lab testing results (10 features), and CT scans upon admission were acquired as three input feature modalities. A deep learning model to extract a 10-feature high-level representation of the CT scans combined with clinical and lab testing features was then constructed. Three machine learning models (k-nearest neighbor kNN, random forest RF, and support vector machine SVM) were then developed based on the 43 features to differentiate all four classes: NS, S, H, and V at the same time. Multimodal features provided substantial performance gain from using any single feature modality. All three machine learning models had high accuracy to differentiate the overall four classes (95.4%-97.7%) and each individual class (90.6%-99.9%) on prediction set. This study will supplement and assist as clinical decision support for the current COVID-19 and other clinical applications with high-dimensional multimodal biomedical features.

Biosketch:

Shi Chen, PhD is an assistant professor of health informatics and analytics at both Department of Public Health Sciences and School of Data Science at UNC Charlotte. He received his Ph.D. in medical entomology and operations research from Penn State University, followed by 5-years of postdoc training in the field of infectious disease epidemiology. His current research areas include modeling COVID-19 epidemic from both traditional epidemiological data as well as social media data in different regions across the socio-cultural spectrum (i.e., “infodemiology” and “infosurveillance”), and exploring multimodal data for clinical decision-making. These studies are supported by the North Carolina Biotechnology Center (NCBC), Models of Infectious Diseases Agent (MIDAS) network, and other external funders.

NCCU

Broadcast Link: Seminar

Abstract:

Tobacco plants are mainly used for smoking, which causes about 20% of all cancers and about 30% of all cancer deaths in the United States. This presentation will introduce efforts on glycoengineering tobacco plants to produce asialo-rhuEPO, a non-erythropoietic derivative of recombinant human erythropoietin, establishing its purification system from transgenic leaves, and testing its in vitro and in vivo neuro- and cardio-protective functions in various cell and animal models. In addition, this presentation will  introduce using the project  as a training platform for students to participate in hypothesis-driven biomedical research.

Biosketch:

Jay (Jiahua) Xie, Ph.D., is a professor in the Department of Pharmaceutical Sciences and a principal investigator in the BRITE Institute (Biomanufacturing Research Institute and Technology Enterprise) at North Carolina Central University (NCCU). Dr. Xie received his Ph.D. in biophysics from the Zhejiang University, China in 1996. He conducted postdoctoral research in the Departments of Genetics and Horticultural Science, North Carolina State University. He took a position as senior scientist at Vector Tobacco Research Inc. and was one of the key scientists involved in research on alkaloid metabolism leading to the development of the first nicotine-free transgenic tobacco plants in the world. After joining the faculty at NCCU in 2006, he continued his research on gene cloning and glycoengineering to produce substances of medicinal/biotechnological importance. Dr. Xie also expanded his research to study cell/tissue-protective effects and action mechanism of plant-produced recombinant human protein asialo-rhuEPO in various cell and animal models.

ECU

Broadcast Link: Seminar

Abstract:

As a health care provider and researcher in an applied field, providing a service to the public in conjunction with research data collection is appropriate.  However, it does present unique challenges that require the practitioner/researcher to make difficult choices.  This presentation will share the challenges as well as the positive outcomes of such an endeavor.  The Driving and Community Bootcamp for Teens and Young Adults with Autism Spectrum Disorder Bootcamp started off as a research activity for a group of occupational therapy students in 2015.  However, with each successive year, the bootcamp has contributed significantly to the community and provided a rich source of research outcomes. The process of building this public service will be highlighted as will the identification of potential research strategies and the supports needed for a productive outcome.

Biosketch:

Anne Dickerson, PhD, OTR/L, SCDCM, FAOTA, GSA is a Professor in East Carolina University’s Department of Occupational Therapy and Director of the Research for the Older Adult Driver Initiative (ROADI).  Dr. Dickerson is an international leader in occupational therapy research in areas of older adults, driver simulation, and drivers with autism spectrum disorder, and driver rehabilitation.  She was awarded ECU’s 2018 Lifetime Achievement Award for Excellence in Research and Creative Activity.  Dr. Dickerson leads the NHTSA project in the state of North Carolina for improving older driver safety.  She was awarded the 2020 NC Governor’s Award of Excellence for Service for her work in promoting driving and community mobility skills with teens and young adults with autism spectrum disorder.

Wake Forest

Broadcast Link: Seminar

Abstract:

With the recent advancements in predictive modeling and machine learning in healthcare, availability and accessibility of biomedical data provides an exciting opportunity for researchers. However, since the majority of the clinical data is unstructured, reproducible and scalable informatics tools that comply with regulations are needed. Wake Forest Baptist Medical Center (WFBMC) has implemented a set of tools for researchers to access free text reports, which has been approved by the Privacy Board and the Compliance Office. First, the reports are processed and deidentified with the open-source MITRE tool. Secondly, a Noble Coder based Natural Language Processing (NLP) pipeline extracts concepts and named entities with customizable vocabularies that have been established. The current terminology is constructed by utilizing the Concept Unique Identifier (CUI) codes found in the Metathesaurus of the Unified Medical Language System (UMLS). Lastly, an Open Distro for Elasticsearch tool is deployed to provide the researcher access and search notes prior to IRB approval. Currently, more than a million notes are available for search.

Additional use cases of this platform include the use of the Elasticsearch for the newly developed specimen search tool that links pathology and radiology reports with the available specimens in our Tumor Tissue Biobank. Machine learning based NLP implementations for specific data marts and potential improvement via federated learning paradigms will also be discussed.

Biosketch:

Umit Topaloglu, Ph.D., FAMIA  is the Associate Director for the Center for Biomedical Informatics, CTSA Informatics Program, and the Wake Forest Baptist Comprehensive Cancer Center. His research involves semantic research data frameworks that include standard based data collection, Natural Language Processing (NLP) and Machine Learning (ML). He also focuses on creating research informatics roadmaps, providing strategic planning that encompasses enterprise data warehousing, data governance, and data sharing networks. He serves on the American Association for Cancer Research (AACR) Genomics Evidence Neoplaisa Information Exchange (GENIE) Executive and Steering Committees and recently served as the co-leader of the Oncology Domain Team for the National COVID Cohort Collaborative (N3C).

Duke

Broadcast Link: Seminar

Abstract:

Use of electronic health record (EHR) data to support quality improvement and research is increasing; however, effective use requires standardization of the data and validation within and across organizations. Information models (IMs) are created to standardize data elements into a logical organization that includes data elements, definitions, data types, values, and relationships. To be generalizable, these models need to be validated across organizations. This presentation will describe a methodology for validation of flowsheet IMs; a data-driven approach to validate and refine existing reference IMs with data from multiple health systems. Flowsheet data were analyzed to standardize concepts, definitions, and associated values sets for assessments, goals, interventions and outcomes. The consensus-based IMs for pain, genitourinary, and falls prevention represent actual data about nursing work in EHR flowsheet documentation. The IM process provides a foundation for optimizing EHRs with comparable nurse sensitive data that can add to common data models for continuity of care and ongoing use for quality improvement and research.

Biosketch:

Kay S. Lytle, DPN, RN-BC is the Chief Nursing Information Officer for Duke University Health System. She is responsible for the strategy, development, deployment, optimization, integration, and maintenance of clinical information systems to support clinicians and patient care across the various care settings. Dr. Lytle also serves as a clinical associate for the Duke University School of Nursing and teaches part-time in the MSN health informatics program. She has 28 years of experience in clinical informatics including system selection and implementation of a wide variety of applications ranging from small departmental systems to large, health system deployed EHRs. Dr. Lytle co-leads the Knowledge Modeling workgroup for the Nursing Knowledge: Big Data Science initiative.

UNC-CH

Broadcast Link: Seminar

Abstract:

Understanding information marginalization of teens and tweens – particularly those who are Black, Indigenous and People of Color – is an important part of developing health communications, information-based interventions, and designing systems and services to serve this group.

A 2016 survey of American teens found that Black and Latine/x youth were most likely to report behavior change in response to health information found on the internet (Wartella et al., 2016), and a little over half (55%) of American teens of all races reported getting “a lot” of their health information from their parents. 40% reported getting a lot of information from doctors or nurses, and 25% reported getting a lot of information from the internet. As BIPOC youth are at increased risk for several conditions, when compared to white peers, it is important that clinicians and others who work with these teens understand how they make decisions about safe or unsafe information sources, and what makes information seeking risky or difficult.

This seminar will present Dr. Gibson’s recent qualitative research on information marginalization and information practices of tweens and teens of color in North Carolina, and discuss implications for health communications and information-based interventions with this group.

Biosketch:

Amelia Gibson, PhD, studies health, wellness, and information practices and access in local communities and on the internet. She is particularly interested in the effects of place, space, power, and community on information worlds, information behavior, information needs, and information access. Her current work focuses on information poverty and marginalization, and how intersections of identity, place, space, and social and economic power/privilege influence information access and information behavior among young women of color and people with disabilities.

Dr. Gibson is an assistant professor in the UNC School of Information and Library Science (SILS). She earned her PhD and MLIS from Florida State University in Tallahassee, and her AB from Dartmouth College in Hanover, New Hampshire.

UNC- Charlotte

Broadcast Link: Seminar

Abstract:

The first law of informatics states that data should only be used for the purpose it was collected. However, informaticians can break this law safely when they define a specific secondary use for it, they have enough data to provide context for the analysis and they understand how the data was collected. To illustrate each point, this seminar will look at three secondary analyses of clinical data in the context of learning health systems and the results from secondary analyses of clinical data. It will discuss the impact of analytical assumptions on hypothesis testing, the impact of data aggregation on analysis scalability, and the impact of data provenance on data quality. Ensuring the reliable secondary use of clinical data is a complex task; concrete examples of major pitfalls to avoid will be provided.

Biosketch:

Franck Diaz-Garelli, PhD is an Assistant Professor of Health Analytics and Informatics at the Department of Public Health Science in the College of Health and Human Sciences at University of North Carolina at Charlotte (UNCC). His research focuses on developing methods and generating actionable information to enable the reliable reuse of clinical data in the context of learning health systems. His research seeks to improve care and prevention of chronic diseases such as diabetes, heart disease, hypertension and cancer. He joined UNCC in August of 2019 after completing a K-12 PRIME fellowship program for the National Institute for General Medical Science at Wake Forest School of Medicine and being part of Wake Health’s Clinical and Translational Science Institute. Dr. Diaz holds a PhD in Health Informatics from the University of Texas Health Science Center at Houston’s School of Biomedical informatics. He also holds a specialized Biomedical Engineering degree from Polytech’ Marseille in France.

NCCU

Broadcast Link: Seminar

Abstract:

In this seminar, research from Dr. Xu’s laboratory on the discovery and mechanistic understanding of using small molecules to treat age-related diabetes and neurodegeneration will be presented.  The seminar will focus on the deleterious protein aggregation developed in various age-related diseases such as type-2 diabetes and Alzheimer’s disease and describe the lab’s effort to discover new molecules with the potential to inhibit toxic entities affecting key organs in the body.  Proof-of-concept examples of using a number of screening approaches including high throughput drug screening and the use of comprehensive tools including chemical informatics to gain mechanistic insights for potential drug discovery will be provided.

Biosketch:

Bin Xu, Ph.D., received training from Fudan University in China, earned a Ph.D. in Biochemistry from Case Western Reserve University in Cleveland, Ohio, did postdoctoral research at the Fred Hutchinson Cancer Research Center in Seattle, Washington and established an independent lab at Virginia Tech (VT). He was recently recruited to the Bio-manufacturing Research Institute and Technology Enterprise (BRITE) of NCCU as an assistant professor of Pharmaceutical Sciences to expand neurodegenerative and aging research and drug discovery programs.

ECU

Broadcast Link: Seminar

Abstract:

The prevalence of older adults dealing with multiple chronic conditions (MCC) has been increasing rapidly in the U.S. and their care now accounts for a large proportion of the nation’s total healthcare expenditure. Support for MCC has become a critical need in reducing hospital utilization, improving health outcomes, quality of care, and reducing the financial burden of chronic disease care. This seminar will focus on a 3-phase study that involved the conceptualization and design, development and usability testing of a Comprehensive Digital Self-care Support System (CDSSS) ― named myHESTIA (my Healing Ecosystem for Self-care and Therapeutic Integration for the Aging) ― for older adults with MCC.  The objective of this study was to test whether a comprehensive self-care support system could be developed for those who are dealing with MCC and whether a system that is specifically developed for older adult patients would enable daily capture of self-care data. The results will be discussed.

Biosketch:

Dr. Priya Nambisan is an Associate Professor of Health Informatics & Healthcare Management in the Department of Health Informatics & Administration at the College of Health Sciences, University of Wisconsin-Milwaukee. She is the Founder & Director of the Social Media and Health Research & Training (SMAHRT) lab and Founder, Developer & Researcher at myHESTIA.org (my Healing Ecosystem for Self-care & Therapeutic Integration for the Aging). Her primary research interests are in the areas of public health informatics, personal health information management, online social and emotional support, wellness and healing technologies and data analytics. Specifically, she focuses on the use of new digital technologies (digital platforms, online forums, and social media) and data analytics (prediction modeling and machine learning) in health. Dr. Nambisan received her PhD in Communication from Rensselaer Polytechnic Institute, Troy, NY and did post-doctoral training in Health Informatics at the Center for Health Enhancement Support Studies, University of Wisconsin-Madison.

Duke

Broadcast Link: Seminar

Abstract:

This presentation will describe the data infrastructure that supports the National Patient-Centered Clinical Research Network (PCORnet), a distributed research network that leverages electronic health records (EHR) and administrative claims for observational and comparative effectiveness research.  It will focus on two of the core components – the PCORnet Common Data Model and data curation process.  The presentation will outline the general design principles behind each component, lessons learned and future directions.

Biosketch:

Dr. Marsolo is an Associate Professor in the Department of Population Health Sciences and a member of the Duke Clinical Research Institute (DCRI) within the Duke University School of Medicine.  Dr. Marsolo received his PhD in Computer Science from Ohio State University. His current research interests include infrastructure to support the use of electronic health records (EHRs) and other real-world data (RWD) sources in observational and comparative effectiveness research, and standards and architectures for multi-center learning health systems.  Dr. Marsolo serves on the leadership of the Distributed Research Network Operations Center (DRN OC) of the PCORnet Coordinating Center and serves as faculty lead for the PCORnet Common Data Model and data curation process (quality assessment). He is a co-lead for the network’s Common Linkage Workgroup (WG), which is working to implement a standard solution for record linkage across the network and is also a co-lead for the PCORnet Data WG. Within the Department of Population Health Sciences, Dr. Marsolo serves as the technical faculty advisor for the DataShare™ Shared Resource, and he serves as faculty advisor to the Pragmatic Health Services Research domain within the DCRI.

Duke

Broadcast Link: Seminar

Abstract:

Electronic health records (EHR) data have emerged as an important resource for population health and clinical research. There have been significant efforts to leverage EHR data for research; however, given data security concerns and the complexity of the data, EHR data are frequently difficult to access and use for clinical studies. We developed a Clinical Research Datamart (CRDM) to provide well-curated and easily accessible EHR data to Duke University investigators. The CRDM was designed to (1) contain most of the patient-level data elements needed for research studies; (2) be directly accessible by individuals conducting statistical analyses; (3) be queried via a code-based system to promote reproducibility and consistency across studies; and (4) utilize a secure protected analytic workspace in which sensitive EHR data can be stored and analyzed.

Biosketch:

Jillian Hurst, PhD is the research director of the Duke Children’s Health and Discovery Initiative (Duke CHDI) and an assistant professor of Pediatrics at Duke University School of Medicine. Jillian received her PhD in molecular pharmacology from the University of Georgia for her work on the regulation of lysophospholipid signaling pathways in ovarian cancer and neurodevelopment in Dr. Shelley Hooks’ lab. She then moved to the University of North Carolina at Chapel Hill to work on ubiquitin-mediated regulation of MAP kinase cascades in Dr. Henrik Dohlman’s lab. Jillian next spent five years as science editor with the Journal of Clinical Investigation and JCI Insight, where she adjudicated research manuscripts and developed review articles and series across biomedical research disciplines.

As the research director of CHDI, Jillian facilitates multidisciplinary child health research to identify early life factors that contribute to long-term health and disease, building collaborations and research infrastructure across Duke University. CHDI infrastructure projects include a longitudinal birth cohort (Project HOPE 1000), linkage of cord blood samples to long-term health records, and the creation of an electronic health records datamart. She also co-leads three observational clinical studies to evaluate early life factors that contribute to health and disease, including HOPE 1000, a longitudinal study of mother-infant dyads, and Duke BRAVE Kids, a study of children exposed to SARS-CoV-2.

UNC-CH

Broadcast Link: Seminar

Abstract:

Incidental Findings (IF), defined as a newly discovered mass or lesion detected on imaging performed for an unrelated reason, are common in the emergency department (ED) setting. The recognition of IFs may be an opportunity for early detection of a serious medical condition, including malignancy. Timely and accurate capture of such findings, however, are challenging and often only recorded in free text reports. In this presentation, we will describe the current state of ED workflow surrounding IFs, summarize the literature surrounding IF in the ED setting, and lastly outline the potential benefits and challenges of using informatics-based solutions, including natural language processing, for aiding clinical decision support.

Biosketch:

Dr. Evans is an emergency medicine physician and 2^nd year clinical informatics fellow interested in integrating informatics tools in emergency and acute care research and clinical decision making. He completed medical school at UC San Diego, his master’s in public health at the Gillings School of Global Public Health at UNC Chapel Hill, and an emergency medicine residency at Vanderbilt University Medical Center. His early research has focused on using large administrative and clinical datasets to better understand the recognition of elder abuse in the emergency department and predictors of repeat EMS transports across the state of North Carolina.  As a clinical informatics fellow, his research interests include Natural Language Processing (NLP) and modeling techniques to predict ED occupancy, crowding and triage decisions.

NCCU

Broadcast Link: Seminar

Abstract:

Cyberattacks on healthcare delivery organizations and other clinical environments have become widespread and pose a significant risk to patient safety and patient privacy. Healthcare IT, security, and clinical engineering teams are learning to work together to formulate specialized cybersecurity strategies in order to lower the risk posed to their respective organizations. Any strategy should be predicated upon a firm understanding of cybersecurity essentials: (1) methods in which cyber-attacks are carried out; (2) healthcare-specific cyber vulnerabilities, and (3) cyber-defense frameworks such as the National Institute of Standards and Technology Risk Management Framework (NIST RMF). This session will focus on how to apply the right people, processes, and technologies for cybersecurity protection in Healthcare IT, with a special focus on medical device security.

Biosketch: 

Jonathan Langer is the CEO of Medigate, a cybersecurity company focused on using connected device data to solve some of the largest challenges in healthcare. He co-founded Medigate in 2017 after 14 years in the Israeli Defense Force Intelligence Corps and is very active in the healthcare cybersecurity and security software community. He co-chaired the Biomedical Security Workgroup for the NC Healthcare Information and Communications Alliance(NCHICA). He holds an LL.Bin Law degree from the Radzyner Law School at the Interdisciplinary Center (IDC) Herzliya, Israel.

UNC-Charlotte

Broadcast Link: Seminar

Abstract:  

Over the past century, mathematical models have been extensively utilized in modeling epidemics. The current paradigm is mechanistic compartment models, such as the most commonly known susceptible-exposed-infectious-recovered (SEIR) models. Meanwhile, data-driven models that usually do not rely on specific mechanisms are not extensively explored. We identify some key challenges of mechanistic models to tackle today’s complex epidemiological systems, explore previously neglected supplementary data sources that could be critical to characterize epidemics, and briefly discuss the feasibility and approaches of data-driven methods for epidemic modeling.

Biosketch:

Dr. Chen is an assistant professor of health informatics and analytics at UNC Charlotte. He started his academic career as a graduate research associate in a WHO-led global vaccination evaluation program of measles. He has worked on various epidemiological systems including measles, vector-borne diseases, zoonotic diseases, HIV, and the current COVID-19 pandemic. He also explores how health misinformation, the “digital pathogen”, infiltrates and proliferates on social media, especially during large health emergencies.

ECU

Broadcast Link: Seminar

Abstract: 

This talk will cover findings from recent and upcoming research on social media’s role in health and persuasive outcomes, with a special focus on COVID-19 and vaccination, climate change, and risk behaviors. Dr. Johnson will discuss how past research on climate change beliefs and risk behavior intentions are impacted by social media exposure. She will also discuss how these research trajectories have led to her current focus on vaccination information, beliefs, and intentions. With COVID-19 variants looming and surging, vaccination beliefs and social mediated information may shed light on how health behaviors are shaped in the coming months and years.

Biosketch: 

Dr. Johnson is an assistant professor at the East Carolina University School of Communication, Greenville, NC. Her primary interests are persuasion, strategic communication, health communication, and research methods in graduate and undergraduate teaching. Her research agenda informs her teaching foci. Her research involves persuasion and how persuasive media can be wielded to impact prosocial and health communication. In the realm of persuasion, her research explores how interactive and entertainment media formats impact psychological processing and health behavior. She additionally studies how social media and human presence attributes (e.g., human voice, attractiveness) can persuade health and potential social change (e.g., in the case of climate change research).

Wake Forest

Broadcast Link: Seminar

Abstract: 

Cardiovascular disease (CVD) has consistently been the leading cause of death in the US and is associated with large healthcare resource utilization. This is partially concerning because of the lack of timely treatments, as the diagnosis of major CVD requires high level and costly imaging such as ECHOs and MRIs. This presentation will focus on the use of a low-cost, easy to access, non-invasive and remote application, an ECG within a deep learning framework to identify patients at high risk for heart failure. Such a tool can identify patients, who may benefit from close monitoring with ECHOs and MRIs, by timely diagnosis of heart failure to prevent their CVD from advancing to the more severe and end-stages of heart failure: class C or D.

Biosketch:  

Oguz Akbilgic, PhD, is an Associate Professor Cardiovascular Medicine and the Wake Forest Biomedical Informatics Center, Wake Forest School of Medicine. He is also the Associate Director for Epidemiological Cardiology Research Center (EPICARE). He received his BS in Mathematics and PhD in Quantitative Methods from Istanbul University, in addition to a MS in Statistics from Mimar Sinan University. He conducted postdoctoral studies in machine learning at University of Tennessee, Knoxville and University of Calgary, Canada.  Before joining Wake Forest, he was an Assistant Professor of Biomedical Informatics at University of Tennessee Health Science Center, Memphis, TN and an Associate Professor of Health Informatics and Director for Public Health at the Cardiovascular Research Institute of Loyola University Chicago, Illinois.  As a health Informaticist, Dr. Akbilgic focuses on methodological developments in the field of artificial intelligence and their applications in clinical decision making. He has worked on health informatics applications in a variety of fields including but not limited to cardiovascular disease, surgery, nephrology, movement disorders, and hematology. He has over 100 publications including two book chapter, and has been the PI and Co-investigator for multiple studies.

Duke

Broadcast Link: Seminar (Please note: first 14 minutes of beginning are cut off)

Abstract:

Early detection of autism is an essential first step toward access to intervention which can impact long term outcome. Although autism screening questionnaires are useful, they require literacy and have lower performance with caregivers from minority racial/ethnic populations. Thus, there remains a need for feasible, accurate, and scalable methods for directly observing and quantifying early autism symptoms.  A digital phenotyping application for early ASD symptom detection and monitoring that can be delivered on widely-available devices and uses computer vision analysis to quantify behavior has been developed. This presentation will discuss how this computational approach can be used for scalable, objective assessment of autism symptoms and monitoring outcomes in treatment trials.

Biosketch:

Geraldine Dawson, PhD is the William Cleland Distinguished Professor of Psychiatry and Behavioral Sciences at Duke University, where she also is Professor of Pediatrics and Psychology & Neuroscience. Dawson is the Director of the Duke Center for Autism and Brain Development. Her research has focused on autism early detection, treatment, and brain function. Dawson served as President of the International Society for Autism Research and was Founding Director of the University of Washington Autism Center. From 2008-2013, Dawson served as the first Chief Science Officer for Autism Speaks and was appointed by the U.S. Secretary for Health and Human Services to the Interagency Autism Coordinating Committee. Dawson is an elected member of the American Academy of Arts and Sciences and was awarded the American Psychological Association Distinguished Career Award (Div53); Association for Psychological Science Lifetime Achievement Award; Clarivate Top 1% Cited Researcher Across All Scientific Fields; NIH Top Research Advances of the Year (2007, 2008, 2009, 2010, 2012, 2013, 2015, 2016, 2017, 2018, 2019); Autism Society of America Award for Research Contributions; Autism Society Medical Professional of the Year; and Autism Society Award for Valuable Service. She is a Fellow of the Association for Psychological Science and American Psychological Association. Dawson received a Ph.D. in Developmental and Child Clinical Psychology from the University of Washington and completed a clinical internship at UCLA.

UNC-CH

Broadcast Link: Seminar

Abstract:

Data is crucial for public health programming. It is often in siloes, time consuming to collect, and rarely goes from raw data to insights, however. This seminar will look at the READI system that developed to harmonize data across programs in Central America, sub-Saharan Africa, and other locations working to end the HIV epidemic and ensure that women everywhere have access to integrated maternal, newborn, child, and reproductive health services.

Biosketch:

Amy Finnegan, PhD, is a and adjunct professor of Global Health at the Duke Global Health Institute and a Senior Data Scientist on the Digital Health team at  IntraHealth International. She earned a PhD in Public Policy Studies from the Sanford School of Public Policy at Duke University and a MPS in International Development Policy and Management at the NYU Robert F. Wagner Graduate School of Public Service. She applies emerging digital health technologies and big data methodologies to problems in global health e.g., HIV/AIDS and reproductive health. Her work seeks to improve program implementation, nurture the data science skills of low and low middle income countries (LMIC) professionals and students, and advance the literature on the use of big data and data science methods in global health. She has expertise in R, geospatial modeling, and business intelligence tools (Power BI/Power Query).

NCCU

Broadcast Link: Seminar

Abstract:      

Drug repurposing is an effective approach to rapid drug discovery, which associates FDA-approved drugs to diseases with no known treatment. These predicted associations (hypotheses) are experimentally tested, and the validated drugs can be used for new diseases. The aim of Dr. Zheng’s group’s research focuses on developing and validating new computational methods for drug repurposing. The methods are either based on text mining of biomedical literature (i.e., the Literature-Wide Association Studies or LWAS) to repurpose existing drugs for diseases of interest or are based on mining graph databases that capture drug-target-disease triples in order to predict potential new relationships among drugs, biological targets, and diseases/conditions. In this presentation, Dr. Zheng will present the problem, elaborate on the approaches, and present both published data and some preliminary data in more recent work on combining text embedding and message passing neural network (MPNN) for drug repurposing. 

Biosketch

Dr. Weifan Zheng is Associate Professor of Pharmaceutical Sciences at NC Central University’s Biomanufacturing Research Institute and Technology Enterprise (BRITE). He has a broad background in cheminformatics and biomedical informatics, with specific training in computational drug discovery. As an investigator at GlaxoSmithKline, Zheng carried out various cheminformatics research work. He was awarded a special Award in Cheminformatics at GSK. As a Sr. Research Scientist at Lilly Research Labs, Zheng was involved in chemogenomics research projects to support the design of gene family targeting chemical libraries. He had been a co-investigator on the NIH-supported project, the Carolina Exploratory Center for Cheminformatics Research (CECCR) at UNC-Chapel Hill. He was the PI of an NIH grant on an integrated informatics system for orphan neurodegenerative diseases as well as grants from rare disease foundations. At NCCU, his group has developed a suite of computational drug discovery tools including receptor-dependent QSAR technologies, a shape pharmacophore-based virtual screening tool as well as text/graph mining technologies for drug repurposing. More recently, he has been working on collaboration projects using text mining to uncover new drugs for rare diseases. In collaboration with researchers at UNC-CH and RENCI (Renaissance Computing Institute), he has employed word embedding technology (Word2Vec) to derive models for predicting P53-protein interactions. He is also working on drug repurposing projects based on graph embedding and graph neural network technologies.

UNC-Charlotte

Broadcast Link: Seminar

Abstract:      

A collaborative effort between Dr. Ge’s lab and a team at Duke Radiation Oncology have been working to develop machine learning models that predict optimal dose parameters for radiation treatment planning. There has been significant progress recently in advancing this AI based technology and translating it into effective and efficient clinical applications that routinely bring the radiation planning time from hours to seconds while maintaining and even improving the quality of the plans. Some of the research efforts needed to make the leap from a good predictive model to a viable clinical application will be discussed during this seminar. 

Biosketch:

Yaorong Ge, PhD, is a professor of health informatics in the School of Data Science and the Department of Software and Information Systems, College of Computing and Informatics at UNC Charlotte. Prior to joining UNCC, he held faculty positions in computer science, biomedical engineering, and radiology at Wake Forest University and Virginia Tech-WFU School of Biomedical Engineering and Sciences. Dr. Ge’s research has focused mainly on technology development in the areas of medical imaging and health informatics. His recent projects include the development of data integration platforms for health data science research, machine learning and decision support frameworks for radiation treatment planning, and machine learning models for predicting heart failure risk in cancer survivors. Dr. Ge received his BS in computer science from Zhejiang University, China, and his MS and PhD in computer science from Vanderbilt University

ECU

Broadcast Link: Seminar

Abstract:  

With the advancements in sensor technologies, there are ever-increasing opportunities for understanding the daily behavior of older adults in order to provide in-time care. Older adults prefer to live their daily lives independently. Also, analyzing the daily routines of older adults allows for a reduction in the demand for care from providers as well as a reduction in the burden on the healthcare system through proactive monitoring of their health.  This seminar will shed light on the capabilities of machine learning in analyzing the daily behavior of older adults. We will go through state-of-the-art data-driven methods for analyzing sequences of ADLs (Activities of Daily Living) with the goal of detecting behavior changes and deviations from the norm that can be early indicators of a health issue.

Biosketch:

Kamran Sartipi, PhD, is an Assistant Professor in the Department of Computer Science at East Carolina University (ECU). His expertise is in data analytics and applications in medical informatics and cybersecurity, and has work with  software analytics, intelligent decision systems, pattern-based data analytics, data and concept interoperability, and cloud computing. His research on intelligent systems includes applying machine learning algorithms with integration of data mining techniques. Dr. Sartipi has over 80 scientific publications and supervised more than 30 graduate students (PhDs and Master’s) in interdisciplinary fields and has developed several software tools in system analysis. He maintains a research platform (DataIntel Research, https://dataintel-research.cs.ecu.edu) with the aims of collaborative multi-disciplinary research initiatives, and organizes computer science seminars at ECU (http://www.cs.ecu.edu/sartipi/CSseminar/).

Fateme Akbari is a Ph.D. candidate in Information Systems at DeGroote School of Business, McMaster University, Canada. She has five years of experience as a Data Scientist in the banking industry, where she did research and implementations on financial data sets with the goal of detecting fraudulent transactions as well as preventing customer churn. Her current research focuses on the use of machine learning in the field of information systems, with a particular emphasis on mining system repositories to detect abnormalities in massive datasets. Using cutting-edge machine learning approaches for sequential data, she is currently performing anomaly detection research on the behavior of older adults.

Duke

Broadcast Link: Seminar

Abstract: 

What is the association of clinician sex, use of the electronic health record (EHR), and work culture with clinician burnout?  We completed a cross-sectional study of 1310 clinicians and found burnout to be more prevalent in women, attending physicians, and advanced practice providers. Multivariate modeling of burnout identified local work culture dimensions accounting for 17.6% of model variance compared with only 1.3% variance for EHR metrics. Female sex independently contributed more to likelihood of clinician burnout [OR 1.3 (1.01-1.75) p=0.04] and significantly interacted with work culture domains of commitment and work-life balance.  Our findings suggest that clinician sex and local work culture may contribute more to burnout than the EHR.

Biosketch: 

Eugenia McPeek Hinz, MD, MS, is Associate Chief Medical Information Officer for Duke University Health System.  She is a Physician Informatician who trained in Internal Medicine/Pediatrics at the Cleveland Clinic and received a Master’s of Science in Biomedical Informatics at Vanderbilt University. Since joining Duke in 2012, she has provided leadership and oversight in the implementation and optimization of the Duke EHR.

As an early adopter of Epic roll out at the Cleveland Clinic in 2001, she excelled in extending electronic documentation to support provider efficiency and improve patient care and quality.  She has extended this work at Duke with a focus on EHR usability to support clinical care and research.  Her own research interests include data visualization, effective clinical decision support and physician well-being.  Additionally she has taught and mentored in Clinical Informatics at Duke for medical students, Clinical Informatics fellows, and Master of Management in Clinical informatics (MMCi) students.