Environmental degradation, social changes, economic hardship and disease are interrelated pathways that impact health. Environmental, social and economic data can be used in monitoring systems to predict the impact of diseases on populations. Our current monitoring systems do not use all this data. Without using this data, our monitoring systems have little predictive power (pixel level) and are largely generalized to administrative units (states). 

Our monitoring systems largely rely on the social determinants of health(e.g., socio-economic, demographic, and genetic conditions). However, this data is spatially coarse, infrequently updated, and costly to measure. Frequently updated, publicly available, fine scale data on environmental determinants of health (temperatureprecipitationair quality) are available but are not extensively used in monitoring systems.

Monitoring systems that have related the environment to health have done so primarily in a single static moment and are not dynamic. Without being dynamic the monitoring systems don’t learn from the data they are monitoring and do not keep up with rates and amounts of change. Without monitoring environmental, social, and economic data and how it is changing, transmission pathways and impact of diseases to populations are unable to be accurately predicted at a fine scale.




Unprecedented environmental, social, economic and epidemiologic data is being generated on the SARS-CoV-2 pandemic. Our solution builds predictive geospatial models on disease impact to populations (initially SARS-CoV-2) using environmental, social and economic data as reference. Disease impact to populations (SARS-CoV-2 cases, COVID-19 deaths) is based upon the amount of factors (environmental, social, economic) that the population is experiencing as well as the rate and quantity of change in these factors. GOSIAC uses this data in machine learning to create a Global Open Source Artificially Intelligent Cloud computing (GOSAIC) spatial monitoring system. Over time this system will increase accuracy by learning about what data (environmental, social and economic) is most effective in predicting disease impact. GOSAIC can extrapolate what it has learned to areas with no disease data available or to emerging infectious disease we know little about. 

In near real time GOSAIC can analyze the data and portray the results in the form of non-technical imagery (heat maps) on a central, user-friendly website. In order to portray imagery in near real time, GOSAIC uses cloud computing to interact with linked data hosted on other sites. First GOSAIC uses data about what the population is experiencing to predict how a population may be impacted by a disease (the resistance surface). This creates an initial raster with pixel values of where populations do and don’t exist, as well as the predicted impact of the population to diseases (cases, deaths). The predicted impact data is used with transportation networks (roads, in the case of human-to-human transmitted diseases) for network analysis. This analysis predicts transmission pathways (i.e., most to least vulnerable population from highways to rural roads). As able, other health risk factors (time since exposure, migration of carriers) and disease impact data (social media reports, contact tracing) can be incorporated.

To encourage participation, collaboration and community learning in the development of this system both the linked data and the coding (behind the artificial intelligence) are open source. GOSAIC allows the global community to improve our understanding of how, why, where, when, and what we are experiencing affects our health.