About the AWS Impact Computing Project at the Harvard Data Science Initiative
The AWS Impact Computing Project at the Harvard Data Science Initiative (HDSI) is an alliance with Amazon Web Services (AWS) aimed at reimagining data science and high performance computing to identify potential solutions for society’s most complex challenges.
The three-year effort will support faculty-led research projects across Harvard. Awards to faculty will be made through sponsored research agreements facilitated by the HDSI and Harvard’s Office for Technology Development.
Summary of Opportunity
The HDSI seeks Letters of Intent (LOIs) from faculty interested in pursuing research that will lead to actionable insights into the social determinants of health. Funding of up to $380,000 total direct costs (maximum three-year term) will be made available for a small number of faculty-led projects on this topic, and HDSI encourages research proposals having smaller budgets and shorter duration and will look to fund a mixture of project sizes. The deadline to submit a 500-word LOI is August 25, 2023.
Priority will be given to projects that identify complex, large-scale, or novel computing or analysis needs, and priority will be given to projects having the potential to drive real-world solutions.
Who is Eligible to Apply for Funding?
This program is open to individuals who (i) hold a faculty appointment at a Harvard school; and (ii) have principal investigator rights at that school. (Please note: Harvard Medical School faculty must hold a faculty appointment with PI rights in one of HMS’s Quad-based, preclinical departments).
Faculty may submit multiple LOIs.
Introduction and Purpose
Social determinants of health (SDOH) are defined as societal and environmental conditions encompassing where people are born, live, learn, work, play, worship, and age that affect a wide range of health, functioning, and quality-of-life outcomes and risk. Social determinants of health can be grouped into the following general categories: Economic Stability, Education Access and Quality, Neighborhood and Built Environment, and Social and Community Context.
The purpose of this funding opportunity is to support faculty-led research projects that contribute to our understanding of SDOH by examining social and environmental drivers of adverse health outcomes and health inequities using population level data. Study designs can include but are not limited to observational studies, panel studies, intervention studies, cohort studies, etc. Researchers might leverage different types of data (e.g., electronic medical records, claims data, etc.) and can include environmental factors such as toxicant exposure and the physical built environment as well as social factors such as living conditions, socioeconomic status, and social inequalities. Funding is also available for studies that examine factors of health equity within and across populations.
The following are examples of research questions that are of interest. These are illustrative only and we welcome LOIs that propose areas of inquiry that fall outside those described below:
Understanding underlying SDOHs
- Increase our understanding of the health consequences of exposure to environmental contaminants (e.g., air pollution, water pollution) both in representative populations and in vulnerable sub-populations (e.g., children, URM). Identification of vulnerable subgroups based on the potential co-occurrence of many factors (e.g., age, comorbidity, income, race, etc.).
- Better understand the health consequences of climate change-related exposures (e.g., heat, wildfires, tropical cyclones). Identification of vulnerable subgroups based on the potential co-occurrence of many factors (e.g., age, comorbidity, income, etc.).
- Identify social and environmental factors that lead to inequities in health outcomes with respect to race, social class, access to care.
- Study of factors that may contribute to gun violence, including mental health, accessibility to guns, poverty, crime, etc.
- Study of factors that improve social mobility, increased access to education, etc.
Developing effective interventions
- Increase our understanding of data, analytics, modeling and prediction methods for successful SDOH interventions focused on housing, food/nutrition, transportation, etc.
- Consider methodological implications of individual vs community-level SDOH data to inform intervention.
- Go beyond process outcomes (e.g., number of people screened or receiving services) to capture additional distal outcomes of social needs interventions (e.g., addressing food insecurity) such as health outcomes/conditions, healthcare utilization, and costs.
- Uncover mediating/moderating factors for social needs interventions such as improvement of care personalization, care quality, reduced patient stress, enhanced self-efficacy, reduced provider burnout, etc.
- Examine evidence-based models for prediction of disease risk (e.g., ASCVD, diabetes) and test addition of SDOH factors for enhanced prediction/modeling risk and reporting outputs that can inform clinical decision support.
Proposed research should address significant data and analytical challenges that might include:
- Developing research data platforms where rich data on granular spatial and temporal resolution are harmonized and linked.
- Consider application of privacy-preserving record linkage (PPRL) methods for linking de-identified datasets (e.g., clinical, claims/financial), accounting for risk of re-identification, potential aggregation into categorical data, etc.
Conducting studies on SDOH requires analytical sophistication because many social and environmental factors are related to each other. Therefore, to prevent disease and eliminate inequities we need to understand which factors are key drivers; this requires new methods for causal inference in the context of high dimensional data. In turn, these studies will require leverage of cutting-edge computation in order to contrast hypothetical subpopulations where only one factor varies (say for example exposure to air pollution) while keeping all the other factors constant. Studies should therefore address methodological challenges. Some examples are listed below:
- Development and implementation of Machine Learning (ML) methods to identify vulnerable populations with respect to several co-occurring factors.
- Development of causal inference methods that allow study of the causal effects of several concomitant factors simultaneously.
- Development of methods for causal mediation analyses to examine direct and indirect effects of SDOH on health outcomes and health inequities.
- Development of synthetic data sets that are representative of vulnerable populations and allow validation of new methods for causal inference and ML.
Submission Process and Requirements
Applicants should submit a 500-word LOI by August 25, 2023.
LOIs will be reviewed by a joint AWS-HDSI review panel and successful submissions will be invited to submit full proposals by October 15th. Funding is expected to begin in early 2024. Prior to full proposal submission, HDSI and AWS will confer with faculty leads and potentially suggest revisions to timelines and scope of proposed projects.
How to Submit an LOI
Please submit your LOI as a PDF to datascience@harvard.edu no later than 11:59 p.m. on August 25, 2023 with subject line “AWS SDOH LOI”. Receipt of your letter will be acknowledged within twenty-four hours. If you do not receive an acknowledgement, please email kevin_doyle@harvard.edu.
LOIs must include the following:
- Title of proposed project
- Names and affiliations of submitting team (PI and any co-PIs)
- Statement of real-world problem that the project addresses
- Description of proposed research questions.
- Description of implications for real-world solutions that may arise from the research, e.g., scalable policy solutions. We especially encourage proposals that identify stakeholders/communities that will benefit from the research outcomes.
- Description of any requirements for large-scale data science, compute-driven modeling, simulation, forecasting, and analytics. Projects should identify complex, large-scale, or novel computing or analysis needs.
- Budget and timeline – this should include initial, direct cost estimates only.
Examples of eligible expenses include:
- Personnel, such as postdocs, research staff, graduate students, undergraduate students
- Travel (domestic and international)
- Acquisition of datasets
Graduate student tuition is NOT eligible for funding.
AWS will make available AWS Cloud Credits to support funded research. At this stage, and where possible, please provide an estimate of need.
Please direct any questions to the HDSI Scientific Director, Lawrence Weissbach at lawrence_weissbach@harvard.ed