Hospital Location Optimization Mixed-Integer Linear Programming Model for Timely Access to EVT Treatment

Abstract

Timely access to endovascular thrombectomy (EVT) is critical for improving outcomes in acute ischemic stroke, yet access varies across Canada due to geographic and system-level constraints. This thesis develops a mixed-integer linear programming (MILP) model to optimize the designation of Comprehensive Stroke Centers (CSCs) within existing stroke systems. The objective is to minimize population-weighted time to EVT treatment while incorporating patient routing, inter-facility transfers, time-dependent EVT eligibility decay, and feasibility constraints. Population and travel-time data were used to model access at a granular geographic level, and candidate CSC hospitals were identified through a structured feasibility assessment. Results show that optimized CSC configurations can reduce treatment times by up to 25 minutes and increase the projected number of patients receiving EVT by 7.42%. This framework provides a data-driven approach to improving stroke system design and reducing disparities in EVT access across Canada.