Skip to main content

Virtual Speed Networking with the Analytic Solutions Committee (ASC)

vid_em_mod
Description

Presented January 11, 2018.

The purpose of the event was to stimulate and facilitate constructive communication and collaboration among analytic method developers and practitioners charged with routine public health surveillance, ranging from disease outbreak surveillance to chronic disease burden assessment and disaster response.



The online event included a series of 5-minute presentations by developers and by public health problem owners, with offline arrangements for much wider participation, including Twitter engagement and topical forums on http://www.healthsurveillance.org/forums/. The talks were intended to give a brief background of interests, needs, capabilities, past accomplishment summaries and links, and contact information. The event concluded with a 30-minute discussion on follow-up activities to facilitate the development of a Developer Network.

Presenters

Developers

  1. Daniel Neill, PhD, Computer Science
    • Presentation title: New Directions in Pre-Syndromic and Subpopulation Health Surveillance

      I will describe two important methodological directions for public health surveillance ("pre-syndromic" or "asyndromic" surveillance, and "multidimensional" or "subpopulation" surveillance), as well as novel approaches my lab has developed to address each problem.  First, while our community has made great strides in developing and implementing syndromic surveillance methods, we must also consider how to discover previously unseen disease outbreaks and other newly emerging phenomena of public health interest that do not correspond to existing syndrome categories.  We must also consider how to use rich information from individual cases, rather than aggregate counts, to better detect and pinpoint outbreaks and other emerging health threats that affect a subpopulation defined by any combination of geographic, demographic, and behavioral risk factors.  I will discuss our current collaborations, developing a pre-syndromic "safety net" for surveillance of free-text ED chief complaint data, and performing opioid and overdose surveillance at the subpopulation level, and hope to connect with additional potential partners for these efforts.

    • Presenter information: Associate Professor of Information Systems, H.J. Heinz III College, Carnegie Mellon University; Director, Event and Pattern Detection Laboratory, Carnegie Mellon University; Visiting Professor of Urban Analytics, Center for Urban Science and Progress, New York University

      Prof. Neill is a computer scientist and data scientist who has spent the past 15 years researching novel statistical and computational methods for heath surveillance.  He has developed numerous methods for earlier and more accurate detection of disease outbreaks, and these approaches have been incorporated into multiple real-world disease surveillance systems.  He has been an advisor to the ISDS Board of Directors and Scientific Program Chair of the ISDS Annual Conference, received an NSF CAREER award, and was named one of the "10 researchers to watch" in the AI field.

  2. Nick Hinkley, MS, Applied Data Science, BS, Microbiology
    • Presentation title: Text Classification of HL7 ELR and On-disk Parallel Computation for Case Linking/Deduplication

      HL7 ELR standards are not always as standardized as we would like or often providers are trying to communicate a message that is not easily expressed by standardized coding systems. Using a simple bag of words representation with an SGD/SVM classifier allows for the high classification accuracy of raw HL7 messages to roughly 60 disease condition classes.

      Historical and disparate public health datasets can become troubling when trying to migrate to new web database applications for case investigations, surveillance, and case linking. Leveraging on-disk parallel computation ISDH was able to calculate similarity scores between siloed HIV, HCV, and HBV datasets in order to get accurate co-infection information along with finding record duplicates on over 10 years of data.

    • Presenter information: Health Informatics Epidemiologist, Indiana State Department of Health

      I worked for roughly 3 years as a clinical microbiologist at Indiana State Department of Health’s Reference Bacteriology/Mycology Laboratory. During my time at the lab I grew increasingly interested in bioinformatics and data analysis which led to the pursuit of a MS in Applied Data Science. I currently work as a Health Informatics Epidemiologist  monitoring quality and issues for HL7 ELR feeds, as Indiana is roughly 94% ELR for lab test reporting.

  3. Maneesha Chitanvis
    • Presentation title: Web-based Biosurveillance Decision Support Tools: AIDO and RED Alert
    • Presenter information: Biosecurity and Public Health (B-10), Bioscience Division, Los Alamos National Laboratory
  4. Patrick Finley
    • Presentation title: Linking data analytics developers and public health practitioners
    • Presenter information: Patrick Finley is a Computer Scientist with the Complex Systems group at Sandia National Laboratories.   Pat models disease processes and develops data analytics to address global health concerns.
  5. Albert Park, PhD
    • Topic: Natural language processing and social media data
    • Presenter information: University of Utah

Public Health Practitioners

  1. Lori Kennedy, MSPH
    • Presentation title: Colorado Center of Excellence in Refugee Health: National ICD Perspectives Wanted

      Colorado is building a refugee health surveillance warehouse made up of data submitted by Colorado and network partners (states, counties, clinics) from around the U.S.  We are standardizing disparate data for public health purposes, and understand that ICD codes are applied differently by clinic, site, state. We have researched a variety of methodologies, and are seeking out insights from others who may have experience with this kind of a project. 

    • Presenter information: Refugee Health Unit Manager, Colorado Department of Public Health and Environment

      Lori Kennedy is a public health epidemiologist with nearly decade of experience, the majority of which has been as a surveillance epidemiologist in communicable disease.  She's a SAS programmer, with crossover into SQL, and currently oversees the Colorado Refugee Health unit within the Disease Control and Environmental Epidemiology Division. She is a co-PI on one of two funded national Centers of Excellence in Refugee Health.

  2. Mamadou Ndiaye
    • Presentation title: Drug Overdose Classification with Natural Language Processing and Deep Learning
    • Presenter information: Epidemiologist, Washington State Department of Health
Referenced File