A data-driven art installation depicting patient activity across a network of health care providers. The Garden blossoms and reacts in real-time as patients are seen by their caregivers across hospitals and clinics throughout the Adventist Health network.
I was the technical and development lead, responsible for designing a reactive garden in code, building servers to handle incoming streaming patient data, and conducting exploratory data analyses to define the scope of variables at our disposal.
Adventist Health had an intriguing proposal for a mediawall in their new headquarters outside Sacramento, California: Build an abstract, data-driven piece that highlights their commitment to providing patient-focused, top-quality healthcare to communities up and down the west coast. Under no circumstances was this to be a traditional data visualization. Adventist Health could provide streaming data of patient activity across their network of hospitals and clinics; it was up to us to come up with how to represent that.
We added an additional challenge: Since this would be in an administrative building, could we build something that would serve as a reminder to everyone there that the work they do each day translates to real people getting real care?
The process started where any good data-driven project should: understanding the numbers. Adventist had upwards of 15,000 data records being generated each day, every time a patient's record was being acted upon in their database. And while no private health information was shared, there were still a number of variables available, like location, type of care, and information about the provider.
We needed to know what variables to use, and how to map those variables to something visual. Moreover, we needed to make sure that whatever streaming data we used needed to be dense enough throughout the day that the visualization would never be empty, but not so dense that it would overwhelm viewers. We started by exploring different variables and working with sample data to see how dense the data was throughout the day and week, and across locations.
For the design, we explored the metaphor of a garden to represent the relationship between Adventist and their patients and communities. In addition, a garden also presented the opportunity to make something that was colorful and dynamic, evolving over the course of the day and throughout the year.
Once we landed on the garden idea, we immediately started prototyping in code to design flowers that could be produced algorithmically. Each patient that appeared in the data would be represented by a flower; variations across flowers would signify different aspect of that patient encounter (e.g. patient age, type of visit) Providers would be represented by butterflies, and would interact with flowers (patients) as determined by the incoming data.
For an additional layer of meaning, we divided the garden into plots that each represented a specific location in the hospital network. Viewers could, at a glance, see which hospitals and clinics were the busiest by checking how dense each location's garden plot was. To make it dynamic, we set the camera on an animation loop to slowly pan and zoom to each location, and added lighting to the scene that tracked the position of the actual sun relative to the mediawall throughout the day.
We traveled to Roseville, CA for the final installation, connected the garden's server to the live stream of patient information, and watched the first flowers begin to bloom.