Last month, a few of our engineers and data scientists chose to spend their professional development day participating in Hospital IQ’s 7th Hackathon. In addition to having a ton of fun, three self-organizing teams ended up with some pretty awesome results:
- Team Cara: Hospital readmission predictions
Team Burt Reynolds: Regional surveillance
Team Raptor Strikeforce: Tracker of financial savings
The concept of a hackathon seems simple: be scrappy, be innovative, and in a single day, push a hard-to-transform idea into working software. In practice, however, the time pressure can become pretty intense. Members of a team must carefully plan how to spend their time and leverage each other’s expertise. How will the work be divided? How much time should be delegated to designing, developing, integrating, and testing? Will there be time to try that new mapping library or modeling technique? Essentially, many of the new product development process questions still apply, but at an accelerated pace. Teams must judiciously choose where to be scrappy or “hack” together functional pieces to ensure they reach the finish line by the end of the day.
Our teams sourced ideas from the entire company and went through the selection process, looking for the most interesting idea. No topic was considered off-limits; an idea that has no fit with our platform, or even the healthcare space, is perfectly acceptable since the primary goal is “binge learning.” As it turns out, though, all three teams ended up choosing ideas that fit will with interest to our existing hospital and health system clients.
Hospital Readmissions Predictions
No hospital wants to see discharged patients readmitted, and Team Cara (named after one of our pets) wondered whether it would be possible to predict which patients might end up being readmitted before they are even discharged. Such forewarning would arm discharge nurses and care managers with additional insight, and perhaps encourage additional oversight and care coordination, to reduce that risk of readmission.
The team took data already on hand in the Hospital IQ platform, and, using a patient-specific machine learning framework previously built by the data science team, built a promising predictive model. For each patient currently in the hospital, the model assigns a score indicating the likelihood of readmission. The model’s initial accuracy already looks promising with 75% AUC (area under the curve).
Showing data as layers on a map can be incredibly useful: take the Hospital IQ COVID-19 dashboard as an example. Team Burt Reynolds wanted to try integrating maps into our platform’s pivot table infrastructure, offering a way to plot a metric of interest arranged by latitude and longitude coordinates using the leaflet.js library. For their proof of concept, they chose transfer center data and highlighted which affiliates were admit sources and at what volumes. The results show transfer cases in a whole new light: it becomes quite clear which geographies most patients are drawn from, and where there are opportunities for growth.
Tracker of Financial Savings
Team Raptor Strikeforce endeavored to showcase something critical to our clients: the return on investment (ROI) seen when using our platform. The team built an interface to customize various inputs into the financial models, such as average margin per elective procedure, and then used those inputs to track changes to the financial health of the hospital over time. These visualizations, intended to be specific to perioperative, inpatient, and staffing settings, tell a compelling story of how operational effectiveness (OE) initiatives, and an investment in the Hospital IQ platform, are paying off. Spoiler alert: it’s pretty easy to spot where adoption of Hospital IQ picks up!
Coming to a Hospital Management Platform Near You
The innovative results from these teams are too good not to share with the world, so we plan on incorporating all 3 projects into Hospital IQ’s hospital management software. We’ll have to replace the duct tape with bullet-proof engineering practices, of course, but we still anticipate all of these features will be available to our clients by the end of the year.
The great work from our team shows that, even on an “off” day where engineers and data scientists can work on anything, they find a way to use their interests–be it on predictive modeling, mapping, or financial calculations–to improve hospital efficiency. And have fun learning along the way!