Machine learning in patient flow: a review.

Area of Science:

• Healthcare Operations Research
• Applied Machine Learning
• Health Informatics

Background:

• Patient flow is a critical challenge in healthcare systems, impacting efficiency and patient outcomes.
• Existing approaches often lack integrated strategies for managing patient movement across different healthcare settings.

Purpose of the Study:

• To review and synthesize the applications of machine learning in improving patient flow within healthcare.
• To identify key areas and challenges in applying machine learning to patient flow management.

Main Methods:

• Systematic review of machine learning applications for patient flow.
• Decomposition of patient flow into four subcategories: demand prediction, transfer logistics, inpatient resource allocation, and length-of-stay/discharge prediction.

Main Results:

• Machine learning offers significant potential for planning, improving, and aiding patient movement through healthcare services.
• Effective patient flow management benefits from both holistic (institutional-level) and granular (patient-level) analytical approaches.
• The development of shared datasets is crucial for benchmarking and advancing research in this field.

Conclusions:

• Machine learning for patient flow is an emerging field with a need for tailored algorithms.
• Future research should focus on transferable algorithms across multiple hospitals and dynamic, real-time decision-making tools for clinical staff.