A Bayesian time-series modeling framework for analysing monthly maternal health service utilization in Ghana: Evidence from antenatal care indicators

Maternal health service utilization remains a critical determinant of pregnancy outcomes, yet temporal fluctuations in antenatal care (ANC) indicators in Ghana continue to limit service continuity and planning. Despite the availability of routine monthly health data, these indicators are rarely analysed using predictive statistical models capable of quantifying uncertainty. This study applied a Bayesian time-series modeling framework to examine trends, seasonal variations, and short-term forecasts of key maternal health indicators including ANC registrants, total attendances, fourth-visit completion, 36-week reviews, and TD2+ immunization using monthly data extracted from the District Health Information Management System (DHIMS2). Descriptive trend analysis, correlation assessment, and Bayesian autoregressive (AR (1)) modeling were conducted. Posterior simulations were used to generate probabilistic forecasts and compare temporal dynamics across indicators. Descriptive trends showed moderate stability in ANC initiation and attendance but a progressive decline in continuity indicators such as fourth-visit completion and 36-week assessments. Correlation analysis revealed strong linkages among early ANC indicators but weak associations involving TD2+, indicating gaps in immunization integration. Bayesian forecasts indicated stable short-term trends in ANC attendance and 36-week reviews, slight declines in ANC registrants, and gradual improvement in TD2+ uptake, though with higher uncertainty. The findings demonstrate inconsistencies in the ANC continuum, particularly from initiation to later pregnancy stages, and highlight the utility of Bayesian modeling for capturing uncertainty and improving maternal health service forecasting. Integrating Bayesian analytical tools into routine health monitoring can enhance early detection of service declines, optimize resource allocation, and strengthen maternal health decision-making in Ghana. Keywords: Bayesian Time-Series, Maternal Health, Antenatal Care, Forecasting.