2026 NexStratus Team
Flagging Operating Room Supply Risk Before It Becomes a Delay
At A Glance:
Capstone Sponsor: NexStratus
Faculty Advisor: Jay Winkeler
Capstone Team Members: Griffin Bodziak, Kate Chittenden, Nolan Crumley, Austin Knapp, Sean Sweeney
Solution Summary: Built the T-72 Supply Chain Smoke Detector, a predictive model and dashboard that scores and flags operating room cases at risk of supply and readiness constraints early enough for hospital staff to act before delays occur, which the team estimated at a net $80,300 against $5,200 in project cost.
The Challenge
Hospitals lack real-time visibility into operational risk, and supply and case-readiness data problems are typically discovered late, producing avoidable delays, inefficiencies, added cost, and revenue leakage. NexStratus asked the student team a focused question: what warning indicators would identify upcoming operating room supply constraints early enough to prevent delays, maintain case flow, and avoid lost surgical revenue? Answering it meant turning scattered hospital signals into one risk indicator that supply chain and procurement teams could act on before a case reached the operating room.
The Solution
The student team worked through four sequential objectives — build a model, gain insights, develop a tool, and integrate it into the NexStratus ecosystem. An initial dataset proved too predictable to reflect real hospital complexity, so the team expanded the data and optimized the model, comparing logistic regression, a neural network, and an XGBoost gradient boosted tree before selecting the boosted tree for its ability to separate critical from non-critical cases. Feature importance analysis identified average case item count and average reserved quantity per procedure as the strongest drivers of readiness risk.
Those predictions power the T-72 Supply Chain Smoke Detector, a dashboard combining the predictive model, risk scoring with flagged cases, and an AI layer. Running on a synthetic demonstration dataset, its overview screen scored 7,676 cases across four risk tiers — 1,196 critical, 1,036 high, 2,980 medium, and 2,464 low — and flagged the 2,232 critical and high cases, 29.1% of the total, as needing attention, with cardiothoracic surgery the largest focus area. A risk predictions view drills into a single day of 1,449 cases, 539 of them scoring 7 or higher, at an average risk score of 5.6 out of 10; model performance and AI compatibility tabs surface accuracy metrics, confusion matrices, and a sanity check on the tool itself.
The Impact
The team valued the project at $85,500 against $5,200 in cost — a net project value of $80,300, with a payback period it calculated at 2.3 weeks of the 38-week engagement. Those are the team's own estimates, modeled on synthetic operating room data rather than realized hospital results. The strategic return runs further: staff time returned to client relationships, a stronger NexStratus product portfolio, and a framework that sharpens as real operating room data replaces synthetic records. The team was candid that the deliverable, though it ranks cases by risk accurately, is not yet production ready, and recommended a pilot launch, threshold optimization, continued feature engineering and data audits, and a hybrid operational view.
2026 NexStratus Team Public Presentation