We come from the field of continuous quality improvement, which requires a gritty, action-oriented mentality. You have to be comfortable “failing toward progress,” understanding that you must define a change, assess its impact, and then use that feedback to design the next version of the change. And you may need to do this multiple times—bringing all those people who are most familiar with the problem along for the ride—before you come to the right solution.
This is a philosophy and methodology we drew upon in redesigning systems of clinical care, but it has deep roots in other industries as well—one of the most famous examples being the Toyota Production System. When an organization is trying to improve the performance of a system, we believe this iterative process is mandatory.
Before we formed Ursa Health, when we were trying to generate the analytics necessary to apply this methodology in a complex, integrated health system, we had a revelation. When we prepared the data to answer the organization’s many complex questions, we took the same gritty, action-oriented approach. We brought together the clinical teams with the data teams. We set up feedback systems and went through multiple iterations. And it worked.
We learned a few other lessons along the way. Because healthcare is complex, clinical and business users may not actually know what question they need answered from the outset, and their clarity is greatly enhanced through this iterative process. And, often, data programmers are the ones who surface incompleteness or patterns in the data that inform the “right” definition of the question. Conversely, data professionals can’t always anticipate the messiness of healthcare data from the outset. Over the course of preparing the data, they will need to make decisions about how to interpret fields from source systems or inclusion/exclusion criteria for the analysis, and these are best informed by the clinician/business end user.
Together, each helps the other succeed in their respective domains through a process that involves creating a rough, first pass at an analysis (e.g., the list of eligible cases or of adverse outcomes), assessing for adequacy (e.g., false positives), adapting the logic based on the feedback, and repeating. At the end of the cycle is knowledge that the organization can trust to guide innovation initiatives.
Our breakthroughs were a set of best practices and programming techniques that elevated the entire innovation team. Analytics consumers were engaged in validating the data transformation decisions, and analytics generators were freed up to facilitate this tight-knit collaboration. We created an organized approach to forming the data and SQL transformation code into reusable components, with each hard-earned asset saved centrally and usable in future reports.
Our development team had no single solution to facilitate this adaptive approach. Instead, we patched together different tools, saved common logic in shared folders, maintained the ETL dependency manually in spreadsheets, and set up the analytic output into the health system’s visualization tools. It worked, but not without a lot of muscle and additional time.
That was the genesis of Ursa Health and our fully no-code analytics development platform. We wanted to make it easier for people with good ideas to bring them to fruition.
Do the harder thing
Embrace thoughtful failure in the service of continuous improvement
Prioritize a calm, respectful workplace
The Ursa team
Interested in our mission?
Do you share our values? Want to be part of a team that’s making it easier for healthcare innovators to do their best work? Reach out to us at email@example.com. And don't forget to visit our News & Insights page to get to know us a little better.