Kevin González Ortega

Kevin González Ortega

Data Science · Psychometrics · AI Evaluation. Fulbright Scholar at Indiana University Bloomington.

PRISM: Panama Research and Integrated Sustainability Model

Team: K. González, E. Aguilar, A. G. Aizprúa, E. Cedeño, J. E. Sánchez-Galán (Universidad Tecnológica de Panamá)

Open the interactive dashboard → · Source code · Paper: APANAC 2023 (in Spanish) (PDF, DOI)

The problem

Food security and child nutrition are central to poverty in Panama, but the data on them was locked in unstructured PDF reports from the Ministry of Health’s (MINSA) Nutrition Program. No geospatial system showed where nutritional problems and interventions were concentrated.

What we built

PRISM (Panama Research and Integrated Sustainability Model) is a geographic information system that brings layers of data about Panama together for interdisciplinary analysis, such as water quality, biological connectivity and biodiversity. Our contribution is its nutrition and human health layer:

  1. Web scraping of MINSA Nutrition Program reports.
  2. Table extraction from the PDFs with Tabula, then consolidation with pandas.
  3. Geospatial join of the indicators with provincial and comarca boundaries using ISO 3166-2 codes, exported to GeoJSON.
  4. Interactive dashboard: choose an indicator and the map recolors each province on a viridis scale, with a tooltip showing each province’s value.

Indicators

The data covers 96 indicators (2003–2014) from the VI (2007) and VII (2013) National Height Censuses of first-grade schoolchildren and MINSA’s 2016 report on nutrition in population groups, including:

  • Census coverage of schoolchildren aged 6–9
  • Mean and median height of 7-year-old boys and girls
  • Prevalence of moderate and severe stunting (height-for-age) by age, sex and urban, rural and Indigenous area
  • Change in stunting prevalence between the 2007 and 2013 censuses
  • First-grade students who do not meet the age requirement
  • Nutritional status of adults by health region

Tools

  • First version (paper): Python (web scraping, Tabula, pandas, GeoPandas) with a Plotly dashboard.
  • Current version: R (sf, dplyr) and a Quarto dashboard with Observable JS, Leaflet and OpenStreetMap, published as a static site.