Data-driven optimization methodology for admission control in critical care units.

Area of Science:

• Operations Research
• Healthcare Management
• Queueing Theory

Background:

• Critical care unit admissions significantly impact hospital performance and patient outcomes.
• Current methods lack a way to integrate patient risk metrics with hospital congestion for selective admissions.
• Hospitals need optimized strategies for managing patient flow in complex care networks.

Purpose of the Study:

• To develop a methodology for selective patient admission to critical care units.
• To incorporate patient health risk metrics and hospital congestion into admission decisions.
• To optimize the allocation of scarce critical care resources.

Main Methods:

• Modeled the hospital as a complex loss queueing network with stochastic patient flow.
• Developed a Mixed Integer Programming model to approximate an optimal admission control policy.
• Optimized a monotonic dual-threshold admission policy for Intensive Care Units (ICUs) and Intermediate Care Units (IMCs).

Main Results:

• The optimized model reduced risk thresholds for admission.
• Weekly average admissions to ICUs and IMCs increased by 37% and 12%, respectively.
• The policy maintained low levels of patient blocking while improving resource utilization.

Conclusions:

• The proposed methodology effectively balances utilization and accessibility in critical care networks.
• It supports personalized allocation of scarce resources to high-risk patients.
• Admission thresholds varying by day of the week show potential benefits for optimizing care pathways.