Modeling the dynamics of time series search interest in autism with implications for public policies

Purpose This study aims to examine and forecast the temporal behavior of Google search interest for the term “autism” in Brazil. The purpose of this study is to investigate the temporal dynamics of public attention and evaluate whether these dynamics can inform the planning of communication and inclusion policies. By linking digital attention patterns to policy timing, territorial outreach and inclusive public communication, the paper contributes to evidence-based strategies for autism awareness and social inclusion. Design/methodology/approach Monthly Google Trends data for Brazil were analyzed using time series forecasting models, including ARIMA, additive and multiplicative Holt–Winters specifications and additive and multiplicative Prophet models. The methodological procedure considered logarithmic transformation, stationarity assessment, differencing where required, residual diagnostics, information criteria and forecast accuracy indicators. The mathematical formulations of the models were retained to clarify the meaning of the stochastic process, smoothing components, seasonal effects and forecast errors. Findings The results indicate a marked seasonal pattern, with recurrent peaks in April and lower levels of relative interest in December and January. Among the evaluated specifications, the additive Holt-Winters model achieved the most accurate forecasting performance across the main error metrics. The ARIMA results indicate limited autoregressive persistence and a pronounced role for seasonal and short-lived innovations in the search series. The weak association between searches for “autism” and “prejudice” suggests little systematic linear co-movement between these aggregate query series; possible explanations for this pattern are cautiously discussed, as Google Trends does not identify individual search motivations. Practical implications The results support the strategic use of digital search data for planning autism-related communication, professional training and inclusive public policies. Awareness campaigns can be better timed around predictable peaks in April, while periods of lower attention can be used to prepare materials, train professionals and strengthen intersectoral coordination. Regional search patterns also suggest the need for territorially sensitive strategies, especially where online information-seeking may compensate for limited access to specialized services. Originality/value The study extends autism-related infodemiology from predominantly descriptive or event-centered analyses to comparative time-series forecasting of a long monthly series in Brazil. By comparing ARIMA, additive and multiplicative Holt–Winters and additive and multiplicative Prophet specifications, the paper evaluates whether recurring patterns of public attention can be forecast with sufficient stability to inform the timing of communication and inclusion policies.