Research Experience & Plans

Raphael Saldanha

Main fields of work and projects

  • Data handling tools: author of ten R packages to handle health and climate datasets and indicators.

{microdatasus}, {brpop}, {tidyrates}, {zendown}, {zenstats}, {nseq}, {rspell}, {brclimr}, {zonalclim}, {opendenguedata}

  • Visualization: malaria and COVID-19 interactive visualization dashboards (IRD-FR and Fiocruz-BR)
  • Analysis and modeling of climate-sensitive diseases (Fiocruz-BR, LNCC-BR, BSC, Inria-FR, IRD-FR)

Health and Climate indicators

  • Standardized methodology to harmonize climate indicators with health data
  • Same time and spatial units (administrative regions)
  • Method of Zonal Statistics
  • Publication at Environmental Data Science journal (2024)

Health and Climate indicators

  • Streamline methods, expansion to other countries and continents
  • Provide regularly updated datasets of health and climate indicators at standardized administrative geographic boundaries (level 2, level 3)
  • Usage of population estimates as weighting factor
  • Build synthetic indicators for water cycle and droughts, warm and cold spells, and extreme events

Climate-Sensitive Diseases modeling

  • Dengue inflicts an important health burden in Brazil
    • 1 million new cases reported just in 2024, 214 deaths, official emergency status declaration
  • Project for dengue incidence forecast on Brazil with machine learning methods
  • Novel ML subsets approach, with cluster based models

Early Warning Systems for CSD

  • Framework proposal to prevent and early detect CSD outbreaks adaptive to regions’ contexts and priorities
  • Aligned with WHO guidelines
  • Close work with health managers and civil society representatives for project usage and tools appropriation
  • Traditional and machine learning multivariate models to forecast incidence for short and medium-time horizons (ARIMA, random forest, XGBoost, LSTM…)
  • Continuous model monitoring and update (MLOps)

Integration with the Global Health Resilience group and projects

  • Toolkits development, maintenance and training. DAGs (directed acyclic graph) to integrate and monitor tools execution and results lifecycle
  • Climate and health data methods and products, basis of different projects
  • ML methods and models for EWS
  • Provide interoperable data, methods and models to different projects, as HARMONIZE, IDAlert, IDExtremes, and E4Warning