123

Purpose: To predict the future number of patients with HIV in Afghanistan and compare the two forecasting models. Methods: An autoregressive integrated moving average (ARIMA) model and a quadratic polynomial Curve-Fitting method were used to fit and predict the number of patients with HIV in Afghanistan. Data from 2010 to 2023 were obtained from the National Statistics and Information Authority (NSIA) of Afghanistan. The coefficient of determination (R²), mean absolute error (MAE), root mean squared error (RMSE), relative error of prediction, mean absolute percentage error (MAPE), the Akaike Information Criterion (AIC) and Bayesian Information Criterion (BIC) were employed to assess and compare the goodness of fit, predictive accuracy, and overall reliability of the two models. Results: Both ARIMA and curve-fitting methods can accurately forecast future trends of HIV cases in Afghanistan. Nonetheless, the ARIMA (5, 2, 2) method had slightly higher accuracy than the Curve-Fitting model (R²: 0.9886 vs. 0.9653; RMSE: 131.33 vs. 228.83). Both models suggested a continued increase in HIV cases between 2024 and 2030, with the ARIMA model predicting 8,979 cases and the Curve-Fitting model projecting 9,198 cases in 2030. Conclusion: This study concluded that ARIMA is the preferred method for forecasting HIV cases in Afghanistan. Accurate forecasting using statistical models can help inform health policymakers in optimizing resource allocation and guiding future public health strategies.