COMPARISON OF FORECASTING DRUG NEEDS USING TIME SERIES METHODS IN HEALTHCARE FACILITIES: A SYSTEMATIC REVIEW

Drug planning is essential to ensure the fulfillment of the right type, amount, and time criteria. Forecasting can be utilized during the planning stage to predict future drug needs. Perfect forecasting is impossible due to uncertainties in various factors, necessitating selecting the best method. This study aimed to identify the optimal forecasting method for healthcare facilities based on the smallest Mean Absolute Deviation (MAD), Mean Square Error (MSE), and Mean Absolute Percent Error (MAPE) values obtained from forecasting results using time series methods like Single Moving Average (SMA), Weight Moving Average (WMA), (Single Exponential Smoothing) SES, Double Exponential Smoothing (DES), and Triple Exponential Smoothing (TES). This research involved a descriptive observational study with retrospective data and adhered to PRISMA guidelines. PubMed, Google Scholar, and Garuda served as the data sources. Nine articles meeting the eligibility criteria were employed. The findings revealed that the SES, DES, and TES methods produced forecasts with MAPE values below 10%, indicating highly accurate forecasting. The MAPE values for the SMA and WMA methods were less than 50%, which is still acceptable. Therefore, the ES methods, particularly SES, are highly recommended for accurate drug planning. Forecasting accuracy factors include data stability, pattern consistency, and smoothing constants. The SES method emerged as the best forecasting method, generating the smallest MAD, MSE, and MAPE values compared to other methods, falling below 10%, reflecting highly accurate forecasting.