Rules Engine Helps to Mitigate Sepsis
Continuing with the previous blog post about user collaboration, in this post I want to share details about a few more of our recent collaborations with Picis customers.
A couple of years ago, one of our users, Dr. Richard Barrow, sent us an article about EHRs (Electronic Health Records) using algorithms for how to detect sepsis early. He suggested that we could do the same with the data that we already collect in the Picis systems.
As we started to investigate the different options, our clinical product team checked criteria used to detect possible sepsis, our engineering team researched possible designs and our customer success managers (CSM) went to talk with our customers.
After conducting a bit of initial research, we understood that sepsis was the first stone to building something bigger: a rules engine. This would be a sepsis alert system where hospitals could establish criteria based on all the data they collect including data coming from devices, documentations from users or data from external providers. The system would analyze all the data, check which rules need to be triggered and alert the user to review the results. These rules could be applied not only to sepsis patients, but also any type of condition with indicator flags. As mentioned in the previous post, Picis software has been collecting data for many years, now our next step is to help hospitals to use this data to help improve the delivery of patient care.
To ensure the new rules engine met or exceeded our customer needs, we started small and vowed to grow the solution based on customer feedback and user experience. As a leading cause of death in hospitals, sepsis has been a major challenge for healthcare organizations to mitigate so it seemed like the logical place to start.
After working on this for months, Picis created an initial prototype that we wanted to validate in a real-world scenario. Dr. Michael Hollis from Bay Pines Veterans Affairs Healthcare System played a key role in building the first product version. Bay Pines' sepsis protocol was tracked via paper until Dr. Hollis agreed to work with us on the construction of the new electronic sepsis alert system. We did our best to match his manual system with our new solution.
User feedback is vital in building a product. In the last few years, we’ve changed our approach in how we can iterate an idea in product development with the help of our users. By doing this we can develop or create solutions that will be closer to the end user’s needs. In the case of sepsis screening and rules engine, user feedback was critical in creating a refined system to avoid alert fatigue.
We had a series of meetings with Dr. Hollis, where we iterated on the product until we could deliver a solution that fit the needs at his facility. During this journey we added the option to allow each hospital to configure the rule without any extra deployment.
Bay Pines Veterans Affairs Healthcare System validated the rules engine in a test environment. After a few changes and adjustments, we released it for production.
After a few months of usage, Bay Pines reported that the new Picis functionality was helping the hospital track possible sepsis situations, so we started looking into the next steps for further developing the solution.
In summary, we took a customer idea and developed it into a solution that is helping sites to improve their day-to-day workflow. We validated the system’s value with the first rule, sepsis, and now it’s installed in hospitals across the world.
If you are interested in collaborating with Picis on product development, reach out to us.
About the Author
Marc Lloses Padilla, Vice President, Research and Development
Marc Lloses brings over 20 years of experience in the engineering world. Prior to joining the Picis team, Lloses worked on Ericsson Mobile R&D and on Siemens Dematic. At Ericsson he was in a pure engineering role, working with the research of new software for mobile platforms. Moving on to Siemens Dematic he was in a more customer-focused role, creating custom solutions for automated storing.