menu

Visualizing Multi-Dimensional Data link

Over the past several months, the Imaging Research Center has increasingly focused its efforts on the problems and promises of data visualization, broadly defined. In particular, we are interested in portraying complex, multifactorial systems, and being able to manipulate that data in real time, often in virtual reality environments. This work began in 2017 with our MapTu software, and has continued with several new initiatives, loosely organized around allowing people to interact with landscapes of interrelated factors.

Screenshot of the MapTu Software in use

The IRC is engaged in a multi-year, multidisciplinary effort to build data-driven visualization systems in immersive virtual 3D spaces. In these spaces people can both work with live data and guide machine learning. We believe that game engines—specifically Epic Games’ Unreal Engine—are the solution to this problem. Game engines are excellent at displaying complex qualitative data (landscapes, objects, characters and their behaviors)’ we want to harness their capacity to show quantitative data with the same granularity and detail.

The VocalMap rendered in paint

The IRC has partnered with Tony Arnold of the Peabody Conservatory of Music and the Resonant Bodies Festival to build VocalMap—a dynamic open-source visualization of the history of vocal music. This comprehensive tool will connect composers, performers and audiences as it displays data about performances, composition, and the interrelationships among multiple pieces. The image above is a painting by IRC director Lee Boot, showing the myriad factors that go into the project. Some are represented abstractly, while others are more literally interpreted. The challenge was to develop a schema that would show all of the factors in understanding the history of Modern Era vocal music and be a generative space where musicians and historians can find new insights and connections

Much of the IRC’s work with complex systems involves analyzing and visualizing healthcare data. We are working on a related artistic visualization for the National Institute of Alcohol Abuse and Alcoholism (NIAAA) to help combat High Intensity Drinking, a particular problem for college students and women over 64. This coral reef-like swirl of logo-glyphs represents factors like peer pressure, and mental health and will ultimately be manipulable in VR. In a similar vein, Lee Boot is part of a working group (including UMBC’s Department of Public Policy and Hilltop Institute) to propose a program involving data collection, analysis and visualization to the Department of Health and Human Services.

Representations of factors involved in High Intensity Drinking

Finally, we are actively collaborating with UMBC’s Hilltop Institute, a health policy research center, to create a visualization of Medicaid data for the state of Maryland that policy makers can interact with in a virtual reality space, using Unreal Engine. This geographic visualization consists of a two dimensional map of Maryland, over which aggregate information about specific health events (for example heart attacks or gunshot wounds) is overlaid in the third dimension, i.e. rendered topographically, with peaks and valleys potentially showing clusters. This kind of dynamic, real-time data presentation would be extremely useful to policymakers allocating resources, distributing care across the state, and in targeting outreach and interventions.

Preliminary visualization of Medicaid data for Maryland

All of these initiatives are united by our belief that the quality of our visualizations, the attention that we pay to aesthetics, give our work greater impact. We believe that allowing policymakers and others to work with data in a tactile way will bring greater insights, and add new utility to this era of big data.