pemr: A Unified R Interface for Managing Personal Exposure Monitor Data in Air Pollution Research
Oscar de León Universidad del Valle de Guatemala

Abstract Personal air pollution monitoring relies on a variety of portable devices, each recording data from different sensors in non-standard formats and data structures, saved to local files or uploaded to web APIs. Tools are needed in R to handle these data in a device- and format-agnostic way. The pemr package aims to provide a unified interface to collect and handle personal air pollution monitoring data.

Data sources

Personal exposure monitors

Air pollution monitoring

• Monitor worn by person
• Standardized setup:
  • Set duration
  • Constant air flow
    (e.g. 1 L/min)
  • Physical separation of particles
  • Filter to capture particles
  • Counting particles in real-time

Current: Device-specific parsing

• Files collected from in-device memory
• Monitoring session metadata saved in file name
• Ad-hoc script to read the file
  • Header with settings information
  • Body with time series of collected data

Issues

• System specific settings
• Code repetition
• Reproducibility more difficult

The role of pemr

# Single function to read data
pemr::read_monitor(
  file = "your-file",
  type = "monitor-type"
)

# Handles each monitor type internally
pemr:::read_em_ecm(file, skip = 24, ...)
pemr:::read_em_upas(file, ...)

# And nested metadata is handled for
# for each monitor type
pemr::collect_data(
  .data = "data-column",
  type = "monitor-type"
)

Further development needed

• pemr handles 4 types
  • RTI MicroPEM
  • RTI ECM
  • PATS+
  • UPAS
• Research monitors
  • Under active development
  • Data structure changes
  • Loss of support
• Consumer monitors
  • Lower precision, but lower cost
  • Smaller and longer monitoring
  • Data web APIs

Future of pemr

• Plugin architecture
  • Handle web APIs
  • Users can define custom data structures

# json
{
  "device_profile": "RTI_MicroPEM_v1",
  "data_source_type": "file",
  "format": {
     ...
  }
}