Data Science / AI

Spatiotemporal Analysis of Childhood Excess Weight in Maranhão

Time-series ecological study covering all 9,609,650 child and adolescent records that SISVAN collected across Maranhão's 217 municipalities between 2014 and 2023. The finding is that the statewide average (20.2%) hides what matters: adolescents drove the entire increase (+0.64 pp/year) and flipped position, the least affected group in 2014 (18.0%), the most affected by 2023 (22.6%), while children stayed flat. The burden is not evenly spread either: a Moran's I of 0.27 confirms spatial dependence, and LISA identifies 25 High-High municipalities in contiguous territories around Bacabal, Pedreiras and Chapadinha. The most consequential technical finding was in ingestion: SISVAN's microdata labels are shifted by one band relative to the official reports, and the intuitive reading would place adolescents at 7.0% instead of 22.6% — the pipeline applies the age-dependent rule and validates against the official reports, aborting on divergence. Submitted to II EANAPS 2026 (SES-MA / UFMA).

Tech Stack

PythonGeoPandasSpatial StatisticsPandasSISVAN

Impact & Stats

9.6M
Records
217
Municipalities
25
High-High Clusters