MIT Sloan Health Systems Initiative

HSI Funds Four Faculty Research Projects from its 2021 Call for Proposals

Funded research aims to impact access to healthy food for diabetics, emergency department operations, data-driven drug discovery, and vaccine hesitancy.

This year HSI was able to fund 60% of requests. We awarded $280,000 across four projects for work that will be done over the next two years. 
Joe Doyle's research on Food as Medicine is continuing. The other three are new projects.  Georgia Perakis focuses on hospital emergency department operations. Vivek Farias's research adds a crucial piece to data-driven drug discovery. And Charles Senteio and David Rand investigate COVID vaccine hesitancy with the ultimate goal of improving willingness.

Georgia Perakis

Even before the COVID-19 pandemic, Emergency Department (ED) congestion had worsened substantially across the US. ED congestion has been associated with increased medical errors, increased risk of in-hospital mortality, and a higher probability of leaving the ED against medical advice or without being seen.  Perakis and team’s research will investigate ways to address not only the important issue of ED congestion, but also ED operations and hospital operations as a whole. They will do this by developing predictive-prescriptive analytic models. These models answer both “what could happen” and “what we can do”. The models will be able to dynamically learn from data and are generalizable to other hospitals. 

The ED at UMass Memorial Hospital underwent a reorganization two years ago and is now separated into three units that process patients differently depending on their acuity index upon arrival. The researchers plan to develop a comprehensive framework based on hospital data to: 

  • Learn, optimize, and prescribe optimal actions within a particular (e.g., the current) ED configuration; and 
  • Potentially recommend different configurations that improve patient outcomes as well as ED and hospital-wide performance metrics. Their goal is to design interpretable and transparent decision-making support tools that optimize performance by combining learning from data with optimization and that could be adopted by physicians and ED staff. 

Vivek Farias

Farias’s work focuses on creating methods to optimize large data sets that have areas of sparse data. His methods allow for making inferences and proving they are valid. This problem of optimizing sparse large data sets and subsequently making appropriate inferences has stumped researchers. This technique is important in, for example, using AI in drug development.

HSI funded some of his prior work in this area, and his platform has already shown promise as a tool to generate proteomic biomarkers at scale for both non-small-cell lung cancer (NSCLC) and Alzheimer’s disease. His work enables more accurate and new conclusions that could not be solved otherwise. 

His newest framework, which builds on the previously funded research, shows promise for creating biologically testable hypotheses that aid in developing new therapies, repurposing drugs, and making some new diagnostic assays more reliable. 

It is an improvement on current practice since it is more data-efficient and thus more powerful. It also permits researchers to investigate more questions than with the current framework because of how it can manage sparse input data. 

Charles Senteio and David Rand

Senteio and Rand’s research explores the connection between cultural beliefs/practices and COVID-19 vaccine hesitancy among Black Americans. There are two phases of this project; this round of HSI funding will underwrite the first part.  

In the first phase of the project, they will use validated research instruments to investigate how attitudes about COVID-19 vaccines vary among Black Americans.  Preliminary work demonstrated that those who were more immersed in African American culture were more skeptical of a COVID-19 vaccine. They will build on this initial work with a larger population to confirm the correlation. 

The insights gained from this initial phase, will form the underpinning for a future project addressing the second phase. Their plan is to design and test tailored messaging to high-risk groups of Black Americans. The hope is that this research can also be used to understand more generally how perceptions many influence health behaviors such as cancer screening, attending primary care appointments, and medication adherence.