1. Health and adaptation of climate change in rural Bangladesh
Bangladesh is recognized as particularly vulnerable to climate change. Projected sea-level rise and more frequent cyclones and floods are expected to have a devastating impact on agriculture, livelihoods and health of its population. Bangladeshis have historically suffered from poor health, due to poverty and water contamination. To adapt to climate change, Bangladesh has developed strategies such as coastal afforestation and reforestation; drinking water provision to affected communities; and construction of water drainage, embankments and roads. However, the health impact of these measures is very little known, particularly during extreme weather events.
This research project will assess the health impact of climate change adaptation measures in Bangladesh. This pilot aims to be an opportunity for researchers across disciplines (such as environment health and geography) and to collaborate on research to improve health. This research will: 1) map the landscape modification by construction and afforestation projects built to mitigate climate change and evaluate their effects on diarrheal disease and mortality among children; 2) examine the effects of the use of safe water techniques (e.g., rain water storage devices, point of use devices, deep tubewells) on diarrheal disease and mortality after extreme weather events; 3) analyze social environment (e.g., education level, socioeconomic status) in the study area and examine the interaction of social environmental factors and climate variability on child health; and 4) recommend strategies for mitigating climate-related events under specific scenarios, focusing on vulnerable populations
2. Predicting human mortality of major causes with advanced analytics
Understanding the dynamics of human mortality is of great importance in the context of a rapidly aging population. Predicting and forecasting mortality helps to make accurate health risk assessments for the general population, to develop strategic plans for preventive healthcare and determine the sustainability of social security systems.
Identifying the magnitude of effects also has important public health and regulatory implications. In this project, we will use California as an example to develop methods and serve as proof of principle for the ideas presented. The results from the project will provide useful information for policy-makers and healthcare professionals.
In this project, we propose novel approaches to predict zip code-level mortality of major causes (such as heart disease, cancers, stroke, diabetes, Alzheimer’s disease, pneumonia and influenza) in California by assembling several machine learning and deep learning algorithms, such as random forests, convolutional neural networks, and recurrent neural networks.
We also identify major factors (such as environmental pollutants and climate factors) contributing to the mortality of specific causes and to predict the mortality of major causes in the near future. The specific aims of the project are listed below:
1) Identify major factors that contribute to the mortality of major causes in California. We will
i) Determine the contribution of environmental pollutants to the mortality of major causes
ii) Determine the contribution of ecosystem services (e.g., greenspace) to the mortality of major causes
iii) Compare and differentiate the contribution from major factors to the major cause-specific mortalities.
2) To develop prediction models for mortality of major causes in California. Specifically, we will
i) Develop machine learning-based models to predict the mortality of major causes, e.g., random forest models and neural network models
ii) Develop deep learning-based models to predict the mortality of major causes, e.g., convolutional neural network models, recurrent neural network models and reinforcement network models
3. Health effects of green space on children autism
Autism spectrum disorders, commonly known as autism, are a group of complex neurodevelopmental disorders typically identified in early childhood. Currently, it is estimated that autism affects over 3 million people in the U.S. Since there is still no cure for autism, children with autism face a great range of outcomes that affect individuals and their families for a long time. As a result, it is highly desirable to discover risk factors and intervention measures. To date, the cause of autism is still unclear. Studies have since linked autism to many environmental risk factors including pesticides, phthalates, toxic waste sites, air pollutants and heavy metals.
Green space (e.g., parks, forests, green roofs, and community gardens), has many benefits to human health and well-being. Green space exerts positive influences on human health through both direct and indirect pathways, such as promoting physical activity and social contact, and mitigating noise and traffic-related air pollution. Contact with green space also affects cognitive function. Given such benefits provided by green space, it is speculated that green space may also have beneficial impacts on children with autism. However, to date, the impact of green space on childhood autism is unclear and epidemiological studies on such an impact are unavailable.
In this project, we hypothesize that green space, particularly near-road tree canopy, has a beneficial impact on autism prevalence by buffering hazards and promoting healthful behaviors. Using information from EnviroAtlas, an interactive mapping tool to connect ecosystem services with human health, we will link fine-scale green space data to autism data from the CDC’s Autism and Developmental Disabilities Monitoring (ADDM) Network. The objective of this study was to examine the association of green space (e.g., forest, grassland, near-road tree canopy) and other factors (e.g., race, socioeconomic status) with childhood autism prevalence in elementary school districts in California. The findings will inform policymakers as to the potential for green space to mitigate childhood autism risk by elementary school district.