Predictive hospital site selection model using machine learning techniques

Abstract

Globally, countries are faced with healthcare challenges that vary from one to the next. While health service delivery challenges are more often seen in countries with a very high HumanDevelopment Index (HDI), inefficient healthcare intervention challenges attract more attention in those with a low HDI. Health systems and infrastructure interventions are major challenges for most countries in Africa. The conventional or traditional approach to situating hospitals has been subjected to the unreliable intuition of experts and perhaps biased by nepotism, favoritism, and tribalism of recognized interest. In this research, we prioritize health factors for hospital site predictions. A hospital is a healthcare intervention infrastructure that should meet the healthcare needs of people where it is located. Many hospital site selection researchers have considered other important factors such as geographical, economic, and socio-demographic factors. However, healthcare factors that will specifically address the healthcare needs of people in that locality have not been mentioned. This paper considers a robust, viable, reliable, and dependable approach to solve the specific problem of selecting the best location for building new hospitals based on the specific healthcare needs of the location for improved healthcare service delivery. We propose a supervised machine learning approach to predict the most suitable sites for building new hospitals. Support Vector Machine (SVM), K-Nearest Neighbours (KNN), Logistic Regression (LGR), and Decision Tree (DT) machine learning algorithm are used to predict the optimal location for hospital site selection on the basis of various attributes used in the dataset. The dataset was extracted from clinical exploratory data using parameters such as reproductive, maternal, newborn, and child health, Infectious diseases, Noncommunicable diseases, Adequate sanitation, Service capacity, and access to essential medicines as a Universal Health Coverage (UHC) standard by World Health Organization (WHO). The model is implemented using Python programming language. The system will improve health systems and infrastructural interventions, thereby enabling efficient healthcare service delivery in developing countries, especially in Africa.