Abstract:

Objective: To analyze the spatiotemporal evolution characteristics of online claims for free contraceptive supplies in Guangzhou and to construct a time series prediction model, thereby providing a quantitative basis for resource optimization and policy adjustment of "Internet-based services" for free contraceptives. Methods: Based on 308 419 online claim records from the Guangdong Province Free Essential Contraceptive Service Management System (2020—2024) and data of the permanent resident women of reproductive age, we conducted multi-time-scale trend analysis and spatial autocorrelation analysis. For the demand forecasting, six time-series models—seasonal autoregressive integrated moving average model (SARIMA), autoregressive integrated moving average model (ARIMA), exponential smoothing state space model (ETS), trigonometric, box-cox transformation, ARMA errors, trend and seasonal components model (TBATS), neural network autoregression model (NNAR), and Prophet model—were compared. These approaches were used to characterize the spatiotemporal patterns of online contraceptive requisitions across the 11 districts of Guangzhou and to project monthly demand for 2025—2026. Results: ①Temporal characteristics. From 2020 to 2024, the annual volume of claims was increased from 0.409 million to 3.702 million units (CAGR=65.2%), showing a step-like climb. The monthly distribution was exhibited seasonality with a "primary peak in March, and secondary peaks in September and December". The claims peaked on Mondays within the week, triggered by official account push notifications from the online platform. ② Spatial characteristics. The per capita claim volume was increased across all districts, during 2020—2024, with Baiyun and Tianhe districts having the highest volumes. The global Moran's I was consistently negative (-0.416- -0.360, P>0.05), indicating no significant spatial clustering, which suggests that the online distribution can break geographical barriers and promote service equity. ③Forecasting models. The highest accuracy of SARIMA (1,1,1) (1,1,1)[12] model was demonstrated, among the six models (MAPE=10.85%). The projected monthly online claims for free contraceptives citywide were expected to increase gradually from 316 800 units in January 2025 to 476 900 units in December 2026. The projected year-on-year growth rates for 2025 and 2026 were 20.14% and 19.01%, respectively, indicating a steady growth across all districts. Conclusions: The demand for online contraceptives in Guangzhou shows seasonality and a sustained upward trend, while demonstrating a pronounced pattern of spatial dispersion. The SARIMA model has a strong adaptability. The spatiotemporal patterns revealed by this study may provide scientific support for the precise supply of contraceptives and the optimization of digital services.

Key words: Delivery of health care, Internet-based service, Spatio-temporal analysis, SARIMA model, Forecasting