GIS for Public Health Research Training Course

Participants can select suitable data models and coordinate reference systems for health studies.

• Vector geometry for health facilities, cases, boundaries and survey locations
• Raster grids representing elevation, temperature, rainfall and population density
• Attribute tables, unique identifiers and relational links between health datasets
• Geographic and projected coordinate reference systems for distance measurement
• Datum transformations and projection checks before combining multiple datasets
• Spatial resolution, extent and scale matched to research questions

Participants can assemble documented spatial datasets from health, population and environmental sources.

• Routine surveillance extracts prepared with dates, locations and case definitions
• Facility registries assessed for coordinates, service type and reporting status
• Census denominators matched to administrative boundaries and reference years
• OpenStreetMap features obtained for roads, settlements and transport networks
• Satellite and gridded datasets selected by spatial and temporal resolution
• Metadata records documenting source, coverage, licence, processing and limitations

Participants can prepare location data for reliable public health mapping and analysis.

• Address standardisation before geocoding patient, facility and laboratory records
• Geocoding match scores reviewed against authoritative administrative reference data
• Duplicate cases identified using identifiers, dates, locations and clinical attributes
• Missing coordinate patterns assessed for systematic reporting or access bias
• Outlier locations checked using satellite imagery, gazetteers and local knowledge
• Quality assurance logs recording corrections, exclusions and unresolved records

Participants can apply confidentiality safeguards to sensitive health location data.

• Direct and indirect identifiers assessed before sharing health location records
• Data minimisation applied to case coordinates, demographics and clinical attributes
• Aggregation rules selected for maps containing sensitive disease information
• Spatial masking techniques including jittering, displacement and geomasking
• Small-number suppression thresholds documented for public reporting products
• Data access agreements defining purpose, custody, retention and disclosure controls

Participants can establish a reproducible GIS project structure for health research.

• Project folders separating raw data, processed data, outputs and documentation
• Naming conventions supporting traceable datasets, maps and analysis versions
• Geodatabases organised with feature classes, tables and coded domains
• Processing history captured through model workflows and parameter records
• Data dictionaries defining variables, units, coding and missing-value conventions
• Map templates configured with standard symbology, layouts and citation fields

Participants can calculate and map health indicators using appropriate denominators.

• Crude incidence rates calculated from cases and population-at-risk denominators
• Prevalence, mortality and case-fatality measures distinguished for mapping purposes
• Age-specific rates constructed using compatible population age-group estimates
• Direct age standardisation applied using a defined standard population
• Rate instability assessed for small populations and rare health events
• Empirical Bayes smoothing introduced for unstable area-level disease rates

Participants can design accurate, interpretable maps for health evidence and decisions.

• Choropleth maps classified using quantiles, natural breaks and equal intervals
• Proportional symbols used for counts where rates would mislead readers
• Bivariate maps comparing disease burden with deprivation or service access
• Colour schemes selected for sequential, diverging and categorical health variables
• Legends, scale bars and north arrows designed for stated audiences
• Map criticism applied to detect visual bias, ambiguity and inappropriate emphasis

Participants can link health records with population, administrative and exposure data.

• Spatial joins assign cases to districts, catchments and enumeration areas
• Point-in-polygon methods link household surveys with administrative indicators
• Areal interpolation estimates populations across changing administrative boundaries
• Dasymetric mapping refines population allocation using settlement and land-cover data
• Exposure zones intersected with population grids to estimate affected populations
• Join validation checks unmatched features, duplicate assignments and boundary errors

Participants can assess geographic access to health services using distance and travel-time methods.

• Euclidean buffers estimate populations within defined facility catchment distances
• Road-network analysis calculates shortest routes between settlements and facilities
• Travel-time surfaces combine roads, terrain, rivers and transport assumptions
• Two-step floating catchment area method estimates service accessibility ratios
• Service catchments compared with administrative planning boundaries and referral patterns
• Accessibility results stratified by poverty, sex, age and rurality

Participants can visualise local concentrations of cases while accounting for analytical limits.

• Point pattern maps distinguish event locations from population distribution
• Kernel density estimation identifies local concentrations of reported health events
• Bandwidth selection tested against transmission patterns and reporting coverage
• Case density compared with population density to avoid misleading hotspots
• Temporal filters separate concurrent outbreaks from accumulated historical cases
• Density surfaces interpreted alongside diagnostic capacity and surveillance completeness

Participants can test whether mapped health values show non-random spatial patterning.

• Spatial weights matrices define neighbourhoods through contiguity and distance
• Global Moran's I tests overall spatial autocorrelation in area-level indicators
• Local Moran's I identifies high-high, low-low and spatial outlier areas
• Getis-Ord Gi* detects statistically significant local hotspot patterns
• Multiple testing considerations applied when interpreting local statistical outputs
• Cluster findings checked against population structure, reporting and data quality

Participants can select and interpret remote sensing variables relevant to health risks.

• Landsat and Sentinel imagery compared for health-related environmental applications
• Vegetation indices derived using Normalized Difference Vegetation Index calculations
• Land surface temperature extracted as a heat and vector-risk indicator
• Rainfall estimates assessed for waterborne and vector-borne disease studies
• Land-cover classes linked with habitat, settlement and exposure pathways
• Cloud cover, seasonality and sensor resolution evaluated before interpretation

Participants can process raster data and produce classified environmental exposure layers.

• Raster clipping aligns satellite layers with defined study-area boundaries
• Resampling methods selected for continuous surfaces and categorical land-cover data
• Supervised classification uses training samples and spectral signatures
• Confusion matrices report overall accuracy, producer accuracy and user accuracy
• Change detection identifies land-cover transitions between selected observation dates
• Raster calculator combines environmental criteria into interpretable risk surfaces

Participants can estimate continuous environmental conditions from point-based observations.

• Inverse distance weighting estimates values from nearby monitoring stations
• Ordinary kriging models spatial dependence through fitted variograms
• Cross-validation compares prediction errors across interpolation methods
• Interpolation uncertainty mapped alongside predicted pollution or temperature surfaces
• Monitoring-site distribution assessed for geographic and socioeconomic bias
• Exposure estimates linked to populations using zonal statistics procedures

Participants can interpret area-level associations without overstating causal claims.

• Ecological fallacy distinguished from individual-level inference and causal evidence
• Confounding assessed using conceptual causal diagrams and covariate selection
• Ordinary least squares residuals inspected for spatial autocorrelation patterns
• Spatial lag and spatial error models introduced for dependent observations
• Geographically weighted regression interpreted cautiously across local parameter variation
• Model results reported with effect estimates, uncertainty and substantive limitations

Participants can develop GIS outputs that support outbreak investigation and response coordination.

• Line lists structured for case onset, reporting date and location
• Epidemic curves linked with mapped case distributions and transmission periods
• Case investigation zones prioritised using clusters, contacts and population density
• Health-facility readiness mapped with staffing, commodities and diagnostic capacity
• Situation maps updated using version control, timestamps and source notes
• Response maps tailored for incident managers, field teams and public communication

Participants can identify geographic inequities affecting health coverage and outcomes.

• Coverage maps compare vaccinated, eligible and zero-dose child populations
• Concentration patterns examined by wealth, remoteness, gender and disability
• Small-area estimates interpreted with uncertainty intervals and model assumptions
• Priority areas ranked using burden, vulnerability, access and intervention feasibility
• Targeting criteria documented to support transparent resource allocation decisions
• Equity findings translated into outreach locations, referral improvements and monitoring indicators

Participants can communicate spatial findings clearly to technical and operational audiences.

• Research map narratives state question, method, finding and decision implication
• Technical appendices document projections, denominators, assumptions and statistical tests
• Uncertainty statements distinguish observed patterns from supported inferences
• Map series structure comparisons across time, location and population groups
• Briefing slides combine annotated maps, indicators and operational recommendations
• Peer review criteria assess accuracy, readability, ethics and decision relevance

Participants can build, review and present a complete GIS public health research project addressing a defined decision question.

• Research question finalised with population, geography, timeframe and decision user
• Source datasets prepared with documented transformations, metadata and quality checks
• Spatial analysis workflow applies rates, access, clustering or exposure methods
• Map portfolio presents findings using suitable symbology, scale and annotations
• Methods review tests assumptions, limitations, privacy controls and reproducibility
• Final presentation defends recommendations against technical and programme review