Scaling Beyond Biomedical: Our First Step Toward Modeling Climate Change as a System
Adam Bly
6.20.2024
As we organize the world's knowledge as one interdependent system, we often think about breadth and depth: Breadth across disciplines that are currently siloed — say, health and environment — and depth across scales that are currently siloed — say, molecular and epidemiological.
Last week, we shared some exciting progress linking biomedical knowledge across scales. Today, we’re announcing an important milestone in our pursuit of breadth: We have started to extract data related to the environment and climate change and include that material in the System Graph.
Climate change is the single greatest driver of complexity in the world — impacting virtually everything in some way, from migration to food security, infectious diseases to housing prices. Our aim is to use AI to represent this dynamic system accurately and comprehensively to help you understand, simulate, analyze, model, and predict its effects better, surface new solutions, and help make the case for investments and interventions.
As of today, you can search for environmental concepts on the latest beta of System and (if you have early access) the System API, and analyze the factors correlated to, affecting and affected by them, all based on peer-reviewed evidence. For example:
Deforestation
Using newly extracted relationships from a study analyzing Landsat-8 imagery, we discover very strong correlations between deforestation rates and three important climate change metrics: land surface temperature, air concentration of sulfur dioxide, and air concentration of nitrogen dioxide. These three metrics are then found in our existing biomedical graph to have downstream impacts on more than 200 health outcomes, including cardiovascular disease, stroke, and asthma (full map here).
Rising Temperature
We’ve started ingesting studies that analyze the correlations between meteorological parameters and climate change metrics. Here we highlight a few paths through which temperature may influence our health:
This study, focused on microorganisms in particulate matter, reported positive correlations between temperature and concentrations of PM10 and PM2.5 particulate matter, which in turn are associated with significant increases in the incidence of asthma, preterm birth and myocardial ischemia.
Temperature and atmospheric pressure are known to be directly related via physics models, but we also capture this relationship as reported via observational data analysis.
A new extraction from this forecasting model study reveals that atmospheric variability is strongly correlated with rainfall prediction in Cape Town, while this study associates seasonal rainfall patterns in Kenya with influenza and hepatitis immunization coverage and success.
Release Risk: You’ll note that environmental concepts currently carry a disclaimer, similar to what you might find on Wikipedia, regarding the completeness of this information. As we expand the source material that we extract data from, further fine-tune our AI, and validate the completeness of these systems against canonical benchmarks (as we have done in health), we will remove this message.
As we embark on this very ambitious project, we are looking to collaborate with climate scientists, climate data producers, and companies and organizations that would benefit from the resulting data. Please reach out if you’re interested in discussing this with us.
Scaling Beyond Biomedical: Our First Step Toward Modeling Climate Change as a System
Adam Bly
June 20, 2024
As we organize the world's knowledge as one interdependent system, we often think about breadth and depth: Breadth across disciplines that are currently siloed — say, health and environment — and depth across scales that are currently siloed — say, molecular and epidemiological.
Last week, we shared some exciting progress linking biomedical knowledge across scales. Today, we’re announcing an important milestone in our pursuit of breadth: We have started to extract data related to the environment and climate change and include that material in the System Graph.
Climate change is the single greatest driver of complexity in the world — impacting virtually everything in some way, from migration to food security, infectious diseases to housing prices. Our aim is to use AI to represent this dynamic system accurately and comprehensively to help you understand, simulate, analyze, model, and predict its effects better, surface new solutions, and help make the case for investments and interventions.
As of today, you can search for environmental concepts on the latest beta of System and (if you have early access) the System API, and analyze the factors correlated to, affecting and affected by them, all based on peer-reviewed evidence. For example:
Deforestation
Using newly extracted relationships from a study analyzing Landsat-8 imagery, we discover very strong correlations between deforestation rates and three important climate change metrics: land surface temperature, air concentration of sulfur dioxide, and air concentration of nitrogen dioxide. These three metrics are then found in our existing biomedical graph to have downstream impacts on more than 200 health outcomes, including cardiovascular disease, stroke, and asthma (full map here).
Rising Temperature
We’ve started ingesting studies that analyze the correlations between meteorological parameters and climate change metrics. Here we highlight a few paths through which temperature may influence our health:
This study, focused on microorganisms in particulate matter, reported positive correlations between temperature and concentrations of PM10 and PM2.5 particulate matter, which in turn are associated with significant increases in the incidence of asthma, preterm birth and myocardial ischemia.
Temperature and atmospheric pressure are known to be directly related via physics models, but we also capture this relationship as reported via observational data analysis.
A new extraction from this forecasting model study reveals that atmospheric variability is strongly correlated with rainfall prediction in Cape Town, while this study associates seasonal rainfall patterns in Kenya with influenza and hepatitis immunization coverage and success.
Release Risk: You’ll note that environmental concepts currently carry a disclaimer, similar to what you might find on Wikipedia, regarding the completeness of this information. As we expand the source material that we extract data from, further fine-tune our AI, and validate the completeness of these systems against canonical benchmarks (as we have done in health), we will remove this message.
As we embark on this very ambitious project, we are looking to collaborate with climate scientists, climate data producers, and companies and organizations that would benefit from the resulting data. Please reach out if you’re interested in discussing this with us.
