EN
Back to the archive

The encyclopedia · R&D & Science · Operational decision · 2013–2016

UT Medical Center scheduled NICU shifts to match doctors' preferences

A mixed-integer program rebuilt neonatal ICU shift schedules around individual physician preferences, lifting preference scores 6.3–8.5%.

University of Tennessee Medical Center

The solution

Hospital units must cover 24-7 shifts, and fairness is usually achieved by dividing shifts evenly by type among physicians — but even splits ignore who prefers nights, weekends or longer shifts.

The team modeled the problem as a binary mixed-integer program that keeps each physician's equality schedule as a baseline and maximizes preference gains subject to coverage and fairness constraints. A hybrid version lets some physicians keep equality schedules while others opt into preference-based ones.

Implemented with a neonatal intensive care group, the schedules improved preference for the opt-in physicians by 6.3–8.5% while maintaining full coverage.

Why it worked

  • Fairness stays anchored as a constraint, so no one loses coverage or equity.
  • Maximizing gains over each person's own equality schedule respects individual baselines.
  • The hybrid model handles groups where not everyone wants to deviate.
  • Optimization finds preference gains no manual scheduling process could.
What it achievedStart from fairness, then optimize preference gainsneat

What can be applied

Fairness can be a constraint, not the objective; anchor on the status-quo fair allocation and let optimization find Pareto-better alternatives no manual process would produce.

Aftermath

The approach was published in Interfaces (2016) and shows how optimization can improve both fairness and individual preference in physician scheduling.

Sources

spotted an error? The archive wants to know.

Related cases