Time series analysis on tuberculosis cases in Kazakhstan from 2014 to 2025

Tuberculosis (TB) is a major cause of morbidity and mortality worldwide. The seasonal autoregressive integrated moving average (SARIMA) model is widely used for forecasting infectious disease incidence, including TB. However, there is a lack of comprehensive research on TB epidemiology and predictions for the future in Kazakhstan. Therefore, this study adopted a time series analysis of 72 monthly observations of reported TB cases in Kazakhstan from 2014-2019. The study used monthly health data from 2014-2019 for TB cases obtained from large-scale administrative records. The methodology involved using the Augmented Dickey-Fuller test for determining stationarity, ACF and PACF for identifying model parameters, and AIC/BIC for selecting the best SARIMA model, which was evaluated based on R2 score, MAE, and MAPE. The modeling dataset comprised monthly TB notification rates from 2014 to 2018, while the forecasting dataset comprised data from 2019. The findings suggest that the SARIMA model demonstrated some efficacy in predicting TB notification rates, although there remains scope for enhancing its precision. The time series was stationary after the first difference (ADF test: t-value of -3.25, P < 0.001) and not a white noise (P < 0.01). Based on ACF and PACF graphs, a SARIMA (0,1,0) (0,1,0)12 model was deemed to be the most appropriate forecasting model, with an R2 score of 0.43, MAE of 89.73 and MAPE of 0.08. To develop effective forecasting models for TB incidence, this study examined monthly TB cases in Kazakhstan from 2014 to 2019 and analyzed the seasonal patterns of TB occurrence. It is important to note that these results are preliminary and necessitate additional validation. Future studies will include correlation analyses between TB cases and meteorological data, using more recent data for TB cases, and evaluating various TB incidence forecasting models. KEY MESSAGES: • A SARIMA model developed for TB in Kazakhstan showed decreasing trend using time series analysis. • Evidence-based strategies for TB prevention & control, and importance of monitoring & forecasting TB incidence highlighted in this study.