Rehospitalization Risk
Making predictive risk information useful at the point of care Predicting that a resident may be rehospitalized is only valuable if care teams can understand the risk early enough and know what to do with that information. This discovery project explored how rehospitalization-risk information could support clinical teams in senior living and post-acute care.
Starting with the the dashboard
The initial temptation with predictive data is to surface a risk score prominently.
Instead, I approached the problem by asking:
Who needs this information? When do they need it? What decision does it help them make?
Research pointed toward the importance of the 14-day period following key care events, where early awareness could help teams identify residents who might require closer attention.
What we learned
A risk score alone wasn't particularly actionable.
Floor nurses and clinical teams needed to understand:
Which residents required attention
How significant the risk was
What factors were contributing to it
Whether something had changed
What action they could take next
That shifted the product question from "How do we display a prediction?" to "How do we integrate risk into an existing clinical decision?"
Design direction
Rather than creating another standalone analytics destination, I explored ways to bring risk information closer to the workflows where clinical teams were already reviewing residents.
The direction focused on:
Prioritizing residents who may require attention
Making risk information easy to scan
Providing context rather than an unexplained score
Surfacing information at the right point in the care workflow
Avoiding unnecessary alert fatigue
Outcome
The discovery helped establish where predictive risk information could provide meaningful value, particularly for floor nurses who could act on changes in a resident's condition.
The project reinforced an important principle in data-driven healthcare design: prediction isn't the product. The decision it enables is.