Incubator Labs in Residence

The Translational Data Analytics Institute is accepting applications for Incubator Labs in Residence, which provide teams with space in Pomerene Hall and staff resources to develop research proofs of concept and extramural proposals while participating in a vibrant intellectual community of faculty, postdocs, students, and industry and community collaborators. All project topics are welcome. Proposals are especially encouraged that relate to climate, environment and sustainability; AI and health; responsible data science; smart cities and mobility; and foundations of data science and AI.

The deadline to apply was Sept. 18; proposals are no longer being accepted.

QUALIFICATIONS - To qualify for a TDAI incubator lab in residence, projects must:

  • Include at least one member who is TDAI core faculty or a TDAI affiliate
  • Include two or more team members who represent distinct disciplines
  • Have an explicit data science component
  • Include target extramural funding mechanisms as a next step

DURATION: One year, with the possibility of renewals.

AWARD COMPONENTS - Incubator lab in residence awards may include:

  • Designated workstations in shared office spaces for graduate students, postdocs and/or research staff
  • Priority access to new reservable collaborative workspaces:
    • Interactive Hybrid Space – 500SF space with touchscreen and ample whiteboard surfaces (12 ppl capacity)
    • The “Dugout space” – 430SF space with 360-degree whiteboard walls (12 ppl capacity)
  • Priority access to reservable meeting rooms for recurring and ad hoc lab meetings (6-16 ppl capacity)
  • Up to 20 staff hours for extramural proposal development for the project
  • Space and up to 20 staff hours for convening activities, including panel/speaker events for up to 175 attendees and seminars/workshops for up to 40 participants. Note: Incubator Lab in Residence awards do not provide funds for participant costs.

Inaugural Labs in Residence (2020-23)

Key Personnel
Dennis Hirsch, Esq. (Law) – Co-PI
Aravind Chandrasekaran, PhD (Fisher College of Business) – Co-PI
Christina Drummond, MA ISTP – Co-PI
Abhinav (Sunny) Hasija (Fisher College PhD student)
Davon Norris (Sociology PhD student)

Project Summary
The Applied Applied Responsible AI Research Lab leverages TDAI’s investment in the Program on Data and Governance by:

  • showcasing active research at Ohio State around emerging data ethics management practices,
  • providing accessible experts who can describe data ethics’ relevance and related emergent initiatives across the OSU enterprise and the national corporate footprint, and
  • providing “open-door” access to OSU data ethics and policy experts actively looking to partner on projects related to data analytics product, process, and management design.

The expected outcomes of the Applied Responsible AI Research Lab’s presence at TDAI include:

  • increased visibility and accessibility of privacy, fairness, and data ethics experts to TDAI research teams, TDAI faculty affiliates, and OSU stakeholders,
  • increased sensitivity and awareness of data ethics among students and TDAI technical research teams during research and development phases, and
  • increased opportunities to share emerging data ethics frameworks, management practices, and regulatory approaches with TDAI visitors and the broader Columbus data analytics community.

Research team members maintain an “open-door policy” when working in residence, and not discussing NDA-covered applied research, to make themselves available to ideate, collaborate and share their knowledge with TDAI faculty and staff.

Key Personnel
Anish Arora, PhD (Computer Science and Engineering) – Co-PI
Rajiv Ramnath, PhD (Computer Science and Engineering) – Co-PI
Matt Lewis, PhD (Design) – Co-PI
Kannan Athreya (Computer Science and Engineering) – Co-PI

Our Lab in Residence is developing in multiple stages a data- and data analysis-centric cybersecurity testbed in Pomerene Hall. The testbed will incorporate Internet of Thing (IoT) devices deployed in a smart campus living lab setting to support a range of education, research and outreach activities on data security. It will consist of a hardware-software system involving beacons, access points, acoustic sensors, radar sensors, software defined radios, and edge computers connected to the OSU network infrastructure and TDAI’s Data Commons. The testbed is to be a core element of an OSU-wide cybersecurity “range,” which we anticipate connecting to the Ohio Cyber Range for statewide users. We will leverage the Aruba-TDAI partnership as part of developing the Data Makery components of the living lab. The innovative edge-computing capabilities developed by our LiR will serve as key infrastructure for SCC&DS RCoP translational community projects, and potentially involve a large number of TDAI faculty, as has already been attempted in our multi-stakeholder funding proposals to the NSF. Finally, testbed use will synergize with activities of Smart@OSU and Smart Columbus.

Key Personnel
Sam Clark, PhD (Sociology); team leader (PI)
Clarissa Surek-Clark (Sociology/English); linguistic analysis/methods
Jason Thomas (Institute for Population Research); methods development and software development
Peter Choi (Sociology graduate student); methods development and software development
Yue Chu (Sociology graduate student); analysis and methods development
Eric Axxe (Sociology graduate student); analysis and methods development

Summary of Work

Verbal Autopsy. Understanding the population-level burden of disease is the foundation for developing and monitoring successful health systems. This relies on identifying all deaths, registering them, and determined their cause. Roughly 60% of deaths globally are not registered or given a cause, and most of these occur in developing countries where health systems are weak. The only feasible solution to providing a cause for deaths in these settings is verbal autopsy (VA). VA is an interview with the family and/or caregivers of a person who has recently died followed by an analysis of the resulting data to determine a likely cause of death. For widespread application of VA, the analysis must be automated in order to limit overall cost, provide timely results, and ensure consistency. Our openVA Team develops statistical methods (machine learning) and related software to ascertain cause of death from VA data and process it in useful ways for health planners:

  • works with the World Health Organization to develop and improve global standards for VA { a standard VA instrument and software,
  • develops new computational/statistical algorithms to automate cause-coding using VA data, and
  • works with the CDC and NGOs to develop software to integrate automated VA methods into national-scale civil registration and vital statistics systems.

Our tools are being used extensively by researchers and by health systems in Bangladesh, Kenya, and Tanzania. Additional countries will very likely begin using our tools over the next year. More

Mortality Modeling. Members of the research team are developing mathematical mortality models for use by the United Nations Population Division (UNPD) to produce their global (205 countries) estimates of age-specific mortality from 1950 to the present, and their forecasts of age-specific mortality from now through 2100. This project will run over the next several years and consists of model development and validation followed by software development to produce production-ready software for the UNPD to use when producing their biannual World Population Prospects report. Additionally, we have a strong collaboration with a team of statisticians at the University of Washington and UNICEF to develop new statistical methods for estimating child mortality at varying degrees of spatial granularity through time using complex survey data. The full team is currently working on producing the official UNICEF estimates of child mortality at subnational scale for Africa.

Epidemiology. Members of the research team conduct a wide range of empirical epidemiological studies in Africa. Recent focuses of this work have been child mortality and the effects of household composition and survival status of parents.

Software. We distribute software using the R Comprehensive Archive Network (CRAN) and Github. All software we produce is open source released under GPL3 or a similar license.

Key Personnel
Mike Rayo, PhD (Integrated Systems Engineering) – co-PI
Courtney Hebert, MD (Biomedical Informatics) – co-PI
James Odei, PhD (Biostatistics)
Megan Gregory, PhD (Biomedical Informatics)
Harvey Miller, PhD (Geography)
Justin Smyer, MPH (Clinical Epidemiology)
Christine Jefferies, RN (Integrated Systems Engineering)
Gregory Metzger, MD (Integrated Systems Engineering)
Morgan Fitzgerald, MS (Integrated Systems Engineering)
Adam Porr, MS, MCRP (Center for Urban and Regional Analysis)

Summary of Work
Our research goals are not only to (1) improve performance in detection of hospital acquired infections and (2) patient decompensation, but to (3) continuously improve the design process integrating subject matter expertise, computational analytics, and visual analytics. The objective of the GeoHAI research is to develop a visualization and assessment tool for Hospital Acquired Infection (HAI), which incorporates geographic data on the hospital and patient-level data from the electronic health record system. We integrate Geographic Information Systems (GIS) with a user-centered design process to build, implement, and evaluate a new computer application which assists hospital epidemiologists in identifying HAI clusters and assessing transmission risk. We expect that incorporation of geographic information into the workflow of hospital epidemiologists will have a profound effect on our understanding of disease transmission and HAI risk factors in the hospital setting, radically altering the workflow and speed of response of infection preventionists and improving their ability to prevent HAI. The objective of the ReComPS research is to develop a set of visualizations and visualization elements whose relative salience change based on how they are directed by both frontline nurses and underlying algorithms being fed continuously by medical device and Electronic Health Record (EHR) data. Across the two projects, the team will be tuning a design process that is influenced by both Human-centered Design and Cognitive Systems Engineering principles and techniques, with the expectation that this increasingly mature process will more reliably produce solutions across multiple settings that produce mutualistic human-machine relationships. With the proliferation of analytics solutions in increasingly high stakes worlds, the impact of a repeatable, reliable process cannot be overstated.