Intelligent Bed Assignment in Emergency Situations: A Hybrid Approach Combining Machine Learning and adaptive Genetic Algorithms (GA)

Managing patient assignment is a major challenge for hospitals at both the strategic and operational levels. It plays a key role in optimizing available resources, especially hospital beds. In this paper, we propose to model the bed assignment problem in emergency situations as a linear programming formulation. The objective function aims to minimize the total cost of assigning, while penalizing patient transfers to reduce unnecessary movements. To solve this NP-hard problem, we adopt a three-step approach. First, we predict patients’ length of stay using a machine learning model, specifically the Random Forest Regressor. Second, based on these predictions, we perform an optimization step that seeks to allocate patients to beds while minimizing the total cost of assigning using a Genetic Algorithm (GA). Finally, a dynamic parameter adjustment strategy is applied to the crossover and mutation rates of the GA to enhance the convergence and quality of the solutions. The proposed method is tested using a set of real data instances taken from a public hospital in Tunisia. The findings represent valuable insights for creating flexible and effective allocation beds plans during emergencies. Experimental results highlight the effectiveness of this approach.