Application of demand forecasting methods as an organizational consulting tool in the surgical center of a private hospital

Abstract

Efficient inventory management in surgical centers is crucial for ensuring quality of care and maintaining hospital financial sustainability. This study aimed to apply and compare time series demand forecasting methods to optimize the management of medical supplies in a medium-sized private hospital located in Caruaru, Pernambuco, Brazil. The methodology adopted a quantitative and applied approach, using consumption data from 22 critical items (Class A) collected between June 2022 and July 2024. A total of 11 models were tested, ranging from traditional methods (Moving Averages, Holt-Winters, ARIMA) to machine learning techniques (LSTM, Prophet, and Ridge and Lasso regressions). The results indicated that the Prophet model achieved the best overall performance, with higher accuracy (lower RMSE) in time series characterized by high seasonality and complexity. In the financial analysis, the proposed model outperformed the current manual management approach, presenting a lower cumulative error relative to the established budget targets. It is concluded that the adoption of advanced quantitative methods reduces the risk of stockouts and unnecessary capital immobilization, offering a practical and replicable solution for hospital logistics. Additionally, the findings demonstrate potential applicability in organizational consulting, by supporting analytical diagnostics and interventions in inventory management, as well as in educational contexts, serving as a basis for the development of case studies aimed at applied teaching in management, administration, and related fields.