Batch 01 · ER triage
Weekend surge wait-time forecasting
-
System failure identification
Saturday evening waits stretch past six hours while Monday clinics sit idle. The pattern repeats every week, but each weekend is handled as an unforeseen event rather than a scheduled one. Staffing for the weekend is set from the same flat average that hides it.
-
Data pipeline diagnostics
Wait time is reported as a trailing weekly mean. Averaging across seven days dissolves a predictable two-day peak into an unremarkable number, so nothing in the report ever crosses a threshold. A threshold that is never crossed cannot escalate, so the system reports itself as healthy.
-
Predictive computational model
Forecast arrivals per hour from the prior eight weeks of same-weekday data, then express staffing as a ratio of forecast arrivals to rostered triage capacity. The ratio breaches before the queue does. Because the input is a calendar pattern rather than live load, the forecast is available days ahead.
-
Workflow integration
Surface the ratio in the Thursday staffing huddle, not on a live wall display. The decision it informs is a roster decision made two days ahead, so the alert belongs in the roster tool. A live display would arrive too late to change anything that matters.