The key objectives of the ‘A One Health Approach’ are:

1. Work at the interface of both Earth science and health science
2. integrate different types of datasets and applications 
3. study the effects of air pollution.

To meet these objectives we developed the following:

• We created a pipeline to extract global satellite air quality data at a resolution of 0.1 degrees per month 
• We created a pipeline to extract NHS prescribing data, available from https://openprescribing.net/, at the resolution of individual drugs at individual clinical sites per month 
• We created plots of the association between these datasets, allowing for understandings of how air quality 

This is important for the following reasons:

• There is an established link between air quality and prescription of respiratory drugs (see https://pubmed.ncbi.nlm.nih.gov/24293452/). However, studies in this area have only made use of ground-level air quality sensors. Acquiring this data can be expensive, and coverage by ground sensors is poor in many areas of the world
• By using satellite data, we can assess air pollution health dynamics in sub-urban areas, non-industrial locations and other areas that may have poor connectivity or where ground-based sensors are not available. This air quality metric provides a cursory predictive value indicating a potentially upcoming decline to respiratory health
• By relying on data pipelines instead of static data files, we can quickly run new models to make new observations: for instance, air quality data satellite data could be swapped for greenery data, and respiratory drug prescription data could be swapped for anti-allergy drug prescription data  

How it works:

• Our system relies on publicly available NASA satellite data and publicly available NHS prescription data
• The system is implemented in python and makes use of standard packages used for data comparisons

What we hope to achieve:

• By creating a system that can be used across the world, we can provide public health services with useful data that would otherwise be unobtainable. 
• We can also build predictive models in areas where we have good prescribing data, which can be used to predict need for drugs in areas that are under-supplied and have no current prescribing data. This data is highly valuable to both public health services in countries where air pollution is developing rapidly, but also to pharmaceutical suppliers who can better forecast regional demand.
• By expanding into additional pollutant indices and other common medications, forecasting health outbreaks, particularly in rural areas, becomes simpler, cheaper and faster.