Comparative Study of Time-Series Forecasting Techniques for Demand Estimation of Highest-Cost Essential Medicines: A Case Study of the Central Chest Institute of Thailand

Abstract

Background: Currently, hospitals are allocated limited budgets for the procurement of drugs and medical supplies. Time-series forecasting has become a technique used to predict future medication needs. The data obtained from forecasting can be used to effectively support decision-making in the procurement and inventory management of drugs and medical supplies.

Objectives: To compare time-series forecasting techniques for the demand of highest-cost essential medicines at the Central Chest Institute of Thailand and to propose a forecast for the upcoming year.

Methods: This retrospective descriptive study used secondary data from the Central Chest Institute to select the highest-cost essential medicines through ABC-VEN analysis. The selected medicines were examined to identify patterns and determine the appropriate time-series forecasting techniques. Subsequently, the time-series forecasting techniques were compared using the mean absolute percentage error (MAPE) as the evaluation metric.

Results: The ABC-VEN analysis identified Jardiance® 10 mg as the highest-cost essential medicine in the previous fiscal year. Among the four selected time series forecasting methods; moving average, weighted moving average, single exponential smoothing, and simple linear regression, the moving average method with a 6-period setting achieved the lowest MAPE of 19.38. This method was determined to be the most suitable for forecasting the demand for this medication. The forecast for the following year estimates a demand of 14,435 units, equivalent to approximately 19.46 million baht.

Conclusions: The 6-period moving average method is the most suitable for forecasting the demand for highest-cost essential medicines, such as Jardiance® 10 mg. This approach provides a reliable basis for rational planning and budgeting for drug procurement.