VACLab Papers at IEEE VIS 2021

VACLab-ers present their latest research including the use of counterfactuals to improve inferences from visualizations, population health for COVID-19, and visualization as a behavioral nudge.

This week brings the 2021 version of the annual IEEE VIS conference. VACLab-ers will be busy throughout the week presenting results from a variety of projects exploring a variety of visualization topics ranging from causal inference to population health.

On the first day of the conference, David Gotz will present a paper reporting on a VACLab collaboration with RENCI and others at UNC-Chapel Hill to apply state-of-the-art visual analytics techniques to clinical COVID-19 data. The work, described in our paper Enabling Longitudinal Exploratory Analysis of Clinical COVID Data to help, demonstrates how the VACLab’s Cadence visual analytics system can help researchers quickly identify and explore key risk factors for COVID-19 from large-scale real world clinical data. This paper is included in the proceedings of the annual Visual Analytics in Healthcare (VAHC) event. Authors include VACLab-ers David Borland, Irena Brain, and David Gotz, along with collaborators Karamarie Fecho, Emily Pfaff, Hao Xu, James Champion, and Chris Bizon.

Second, VACLab teammate Alex Rich will give a presentation during the Visualization for Social Good event on Day 2 of IEEE VIS. Alex will describe his work exploring the use of local and personal visualizations of COVID-19 data as a behavior nudge and commitment device. You can learn more about this project in the accompanying paper titled Local, Interactive, and Actionable: A Pandemic Behavioral Nudge. The work was a collaboration between VACLab-ers Alex Rich and David Gotz, and Cameron Yick from DataDog.

Finally, VACLab-er Smiti Kaul will present work from VACLab teammates Kaul, David Borland, and David Gotz, in collaboration with Nan Cao from Tongji University, about the use of counterfactuals to help improve causal inferences drawn by users from visual representations of data. The paper, Improving Visualization Interpretation Using Counterfactuals, will appear in IEEE Transactions on Visualization and Computer Graphics. It describes the use of counterfactual data subsets to contextualize visualized data in order to help users better understand why they may see differences across subgroups. The paper includes results from a user study comparing user inferences between two visual analytics systems: one with counterfactual visualization as context, and one without. The results show that visualizing counterfactual data holds promise as a way to help users make more informed and appropriate inferences about relationships between variables during visual analysis.

References

Smiti Kaul, David Borland, Nan Cao, David Gotz (2022). Improving Visualization Interpretation Using Counterfactuals. IEEE TVCG (To Appear).

Alex Rich, Cameron Yick, David Gotz (2021). Local, Interactive, and Actionable: a Pandemic Behavioral Nudge. Proceedings of Visualization for Social Good.

David Borland, Irena Brain, Karamarie Fecho, Emily Pfaff, Hao Xu, James Champion, Chris Bizon, David Gotz (2021). Enabling Longitudinal Exploratory Analysis of Clinical COVID Data. Proceedings of Visual Analytics in Healthcare.