A tree-based prescriptive analytics framework for contextual surgery scheduling

doi:10.1016/j.omega.2026.103660

A tree-based prescriptive analytics framework for contextual surgery scheduling

2026-09-21

0

PRE

AI

DOI：10.1016/j.omega.2026.103660

Original

Original request for help

Save

Abstract

En 中文

• The first prediction-free prescriptive framework for contextual surgery scheduling. • A solver-free pairwise discrepancy loss trains the tree without re-optimization. • We prove a tighter scheduling-cost bound than classical wait-and-see relaxations. • A loss-compatible branch-and-cut algorithm accelerates sequential daily scheduling. • Our approach reduces average regret ∼ 50% on synthetic and real cardiac surgery data.

Keywords:

Prescriptive analytics

Contextual optimization

Operating room scheduling

Uncertainty

Tree-based machine learning models

A tree-based prescriptive analytics framework for contextual surgery scheduling

Abstract

Journal

Organization

Cited Papers