SARA

EVIDENCE-BASED DECISION SUPPORT

Uruguay

Uruguay — Infant mortality rate

In Uruguay, the analysis focused on the infant mortality rate, an indicator that is already at relatively low levels within the region. The initial reduction target was adjusted because the final model included fewer variables and a smaller number of localities than in other countries, which reduced the scope for simulation. The scenario prioritized poverty, the urban building index, household internet access, beneficiaries of the Tarjeta Uruguay Social, and peri-urban nighttime light radiance.

These variables were selected because they make it possible to represent socioeconomic conditions, urban infrastructure, connectivity, social vulnerability, and territorial dynamics. Compared with other variables, priority was given to those that remained available in the model and could be modified within a plausible counterfactual scenario. The scenario estimated a reduction of 7.36%, reaching a final rate of 5.52 compared with 5.96 in the modeled value.

These variables were considered relevant because they reflect basic household material conditions, energy access, and socioeconomic vulnerability. They were prioritized over other variables because they make it possible to represent concrete changes in infrastructure and household well-being, two key dimensions for interpreting territorial inequalities. The calculated scenario showed an estimated reduction in low birth weight of 0.66 percentage points, equivalent to a decrease of 6.70%.

Uruguay's Dashboard

Work Packages

Data Architecture
  • Identification and standardization of variables across scales; development of indicators
  • Analysis of spatial patterns and factors associated with Food and Nutrition Security pillars
  • Geospatial characterization of community food environments: individual (socioeconomic), environmental (water/food access & availability; physical-climate), political (public policies linked to health and food)
  • Integration of environmental conditions via satellite data
  • Evaluation of vulnerability and resilience to climate change
  • Definition and development of a FNS metric; training and validation of predictive AI models
  • Simulation of future scenarios at 10, 30, and 50 years based on climate change and public policy interventions
  • Build an interactive and dynamic dashboard capable of simulating future scenarios based on climate change and policy interventions

  • Enable decision-makers to explore different intervention pathways and their projected impacts on Food and Nutrition Security

  • Implement a virtual assistant (chatbot) powered by a GPT-style generative model to help users access and interpret dashboard outputs and simulations

  • Execute dashboard models in the background

  • Provide automated responses and personalized recommendations for decision-makers in real time