Environmental Health Sciences · University of Arizona
Mahdi Faraji
What you cannot see is still a dose.
Air<>Orbit<>Climate<>Intelligence<>Action
Connected through Digital Public Health
I — Exposure
The invisible burden
Air reports itself as clean because perception is not tuned to what it carries. The particles that matter most sit furthest below the threshold of sight, and the smallest travel deepest, past every filter the body has, into the tissue where gas exchange happens.
Size decides depth. Coarse material is caught in the head airways, finer material reaches the bronchi, and below a tenth of a micron diffusion carries a particle to the alveoli. Exposure becomes dose without announcing itself.
My work begins at that gap between what is present and what is perceived. It began in workplaces, where the gap is widest and consequences arrive soonest.
Particulate matterOccupational exposureDose
II — Measurement
A reading is not yet evidence
Instruments became cheap before they became trustworthy. A few-hundred-dollar sensor will produce a number every minute for years, and almost none of it is defensible until that sensor has stood beside a reference long enough to reveal exactly how it lies.
Drift, gain, humidity, season. The correction is not a constant but a learned function, and its only honest test is a future the model has not seen, never the comfortable case where training and testing share the same weeks.
This is where machine learning belongs in environmental health: as an instrument, held to the standard of one.
CalibrationMachine learningValidation
III — Resolution
Certainty is a matter of density
Regulatory networks answer at the scale of a county. People live at the scale of a block. Almost everything determining who is actually exposed sits between those two resolutions, and conventional monitoring is too expensive to close the distance.
Move away from an instrument and confidence drains. Density buys it back. From orbit the coverage is total and the resolution coarse. On the ground it is the reverse. The work is in the fusion.
Heat behaves the same way and arrives with the same inequity. Treating temperature and air as one exposure surface is the more honest account of what a person standing outside is subject to.
Sensor networksSatellite retrievalSpatial predictionClimate
IV — Translation
A number nobody uses has done no work
The final step is the one the discipline is worst at. A measurement that reaches a journal and stops there has changed nothing for the person breathing the air it describes. The distance between a validated dataset and a protective decision is itself a research problem.
It is an interface problem and a language problem. The question a resident asks is not the question a model answers. The work is to build what stands between them, grounded strictly in what the network measured and honest about what it does not know.
Here public health, digital infrastructure and environmental measurement stop being separate fields.
Public healthCommunicationApplied AI
V — Profile
I am a doctoral researcher in Environmental Health Sciences at the University of Arizona, in the Mel and Enid Zuckerman College of Public Health, with a minor in digital public health.
I came to environmental measurement through occupational health engineering, first in industry and then in research. Years on factory floors preceded the laboratory, and that order matters. Whether a measurement can carry the weight placed on it was a practical question for me long before it was a methodological one.
My work spans air quality and environmental health, occupational health and safety, ground sensing networks and satellite retrieval, machine learning in exposure science, climate and sustainability, and health innovation.