Uso de metaheurísticas para auxiliar no posicionamento de unidades básicas de saúde

Abstract

The Basic Health Unit (UBS) is the means by which the population has access to the health system; however, it is important that they are accessible to society, possessing a good location. The correct location of these units is fundamental to guaranteeing the efficient access of the population to basic services, especially in contexts marked by limitations of resources and infrastructure. Thus, the challenge arises of finding alternatives that allow determining the location of the UBSs. Given this scenario, the objective of this work is to identify a strategy capable of determining the positioning of the Basic Health Units in a way that can serve the largest number of people in a region. In this work, two metaheuristics were selected: Greedy Randomized Adaptive Search Procedure (GRASP) and Genetic Algorithms in solving a facility location problem, focusing on maximizing the coverage of care in public health services. Defining the ideal location of primary healthcare units (UBSs) is a combinatorial optimization problem, as it involves factors such as the geographical distribution of the population, the limited number of available units, and the distances between demand points. Due to the high computational complexity, the application of exact methods becomes unfeasible for large instances. In this scenario, metaheuristics stand out as efficient alternatives for obtaining good quality solutions in reduced computational time, assisting in decision-making. The results obtained indicated that, for instances with a smaller number of UBSs, both algorithms showed similar performance. As the number of UBSs increased, the GRASP algorithm obtained higher quality solutions, achieving percentage gains greater than 6% and 7% for four and six UBSs, respectively. In contrast, the Genetic Algorithm presented significantly shorter execution times, with speedups greater than 3.6 and 7.4, demonstrating the suitability of each approach according to the time and solution quality constraints