Anonymisation of address coordinates for microlevel analyses of the built environment: a simulation study

Area of Science:

• Spatial epidemiology
• Geospatial analysis
• Public health research

Background:

• Data privacy is a major concern in spatial epidemiology due to the use of exact residential locations.
• Spatial aggregation methods can lead to a loss of spatial variability.
• Exact residential addresses were not permitted for microlevel analyses in the IDEFICS study.

Purpose of the Study:

• To implement a spatial blurring method to anonymize address coordinates based on population density.
• To assess the impact of spatial blurring on microlevel analyses of the built environment.
• To enable data protection compliance while conducting detailed spatial epidemiological studies.

Main Methods:

• Added Gaussian distributed error to address coordinates, with variance dependent on population density and k-anonymity.
• Generated 1000 random point locations, blurred 100 times each, to create anonymized locations.
• Calculated walkability indices using 1km network-dependent neighborhoods for both blurred and original locations.

Main Results:

• Spatial blurring's effect decreased with increasing population density.
• Mean differences in walkability indices were smaller in densely populated areas (≥1500 residents/km²).
• Spatial blurring did not significantly affect the assessment of the built environment in densely populated areas.

Conclusions:

• Spatial blurring enables microlevel investigation of the built environment while ensuring data protection.
• The method allows for individual network-dependent neighborhood analysis.
• Minor effects on walkability assessment were observed, primarily in sparsely populated areas.