Modelo de previsão de demanda do serviço de urgência em um hospital de pronto atendimento

Abstract

In this work, the topic of forecasting demand for emergency services in the emergency department was addressed. The research objective was to propose a demand forecasting model for emergency consultations. Consisting of two academic articles, the first article is a systematic review of the literature on demand forecasting methods with the purpose of bringing together the methods and models available about the concepts currently used in the management of companies related to the consumption and production of products and services. The methodology used is the systematic review of the literature with a qualitative approach in order to give an overview of the dominant methods used in demand forecasting. The literature was mapped to identify the state of science through the available scientific production. Qualitative and causal methods show better adaptation to medium and long-term forecasts, while the analysis of time series indicates that it is more suitable for shortterm forecasts. The second article proposes a model to predict the number of daily visits to an emergency hospital. Data from a 66-month period from a private Emergency Care and Day Hospital located in Rio Grande do Sul, Brazil were used. It is a quantitative modeling research, characterized as empirical normative quantitative, with the purpose of providing the identification of the current stage of knowledge regarding the forecast of demand for emergency consultations in the emergency department and through a probabilistic model to propose a methodology for assertiveness and efficiency gains in the studied environment. Data modeling was performed using Microsoft Excel software for data collection and organization and free software R Version 3.6.2 to build the demand forecast model and statistical analysis. The results suggest the ARIMA model (1,1,4) for adult general practitioners and ARIMA (4,1,1) for pediatricians, considering the best possible adjustment for predictions. Forecasts from 1 day ahead to 14 days of forecast are recommended, with errors close to 10%. In addition, as an alternative demand forecast model, it was proposed to use the chaotic model by the logistic map method for short-term forecasting in emergency consultations, incurring a low error associated with the forecasts. The study demonstrates the importance of using mathematical forecasting models in emergency department care services and as a management tool.