Faculty/Staff Research | Data Science

Research Projects

The following projects reflect the commitment of Mississippi State University's academic departments and research centers to performing relevant basic and use-inspired data science research that addresses the world's most pressing problems.

Off-Road Autonomy

A self-driving vehicle is a digitally operated system that relies on many aspects of data science to work properly. It detects objects (pedestrians, signs, bicyclists, exits) using computer vision techniques that rely on massive amounts of training data. It manages data related to all car systems that come from sensors like GPS, radar, and lidar for navigation; drivetrain sensors for temperature, charge, speed, and acceleration; and even sensors that ensure the human occupant is paying attention in case he or she needs to take the wheel. Yet, driving a car on paved roads with well-organized signs is not the only challenge for autonomy research. In contexts like smart agriculture, military, or rural contexts, vehicles must navigate in off-road settings away from automotive-specific signals like signs and roads. Off-road, industrial, and heavy-duty vehicle automation are among the last frontiers of autonomous mobility. Mississippi State’s Center for Advanced Vehicular Systems (CAVS) is on the forefront of autonomous mobility research, focusing on non-urban environments. With top-rated high-performance computing capabilities and one-of-a-kind vehicle proving grounds, CAVS validates advanced modeling and simulation developments in real-world situations. CAVS also has full-suite capability for autonomous system development, including sensor research, artificial intelligence, and vehicle robotization.

Scientist: Daniel W. Carruth
Responsible Unit: CAVS
Project Information: Autonomous Vehicles

Trends in Northern Hemispheric Snow Presence

Mississippi State researchers partnered with researchers from the University of California at Santa Cruz and the University of North Florida to study global snow cover. Remote sensing data, from sensors mounted on satellites or aircraft, allow scientists to address questions related to agriculture, population, weather, climate, and many other fields using statistical strategies. This research project made use of weekly snow cover extent data gathered by the National Oceanic and Atmospheric Administration (NOAA) using satellite flyovers and presented via the Rutgers Northern Hemisphere Data Map, a pixelated rendition of snow presence. The map provides snow cover data for each pixel – 0 for not covered and 1 for covered. Mississippi State researchers developed mathematical models and constructed statistical methods to estimate trends at each pixel. The novel models account for the 0/1 structure of the data and correlation. Additionally, for the first time, uncertainty quantification for correlated 0/1 time series data is addressed, allowing researchers to statistically test a hypothesis that snow cover is changing. Approximately twice as many pixels are experiencing decreasing snowpack as compared to increasing snowpack.

Scientist: Jon Woody
Responsible Unit: Mississippi State University Department of Mathematics and Statistics

Open-Source Intelligence

Open-source intelligence experts collect and analyze data gathered from sources that are publicly available like internet discussion forums and social media sites to produce actionable intelligence. The Open-Source Intelligence Lab (OSIL) at MSU’s Center for Cyber Innovation (CCI) conducts groundbreaking work using open-source intelligence (OSINT). This work is led by a network of interdisciplinary faculty experts working at the intersection of publicly available information (PAI), data science, and ethics. Using deep faculty experience in this field, combined with state-of-the-art open-source data aggregation and analytics resources, CCI conducts applied research that informs our collective understanding of some of society’s and government’s most pressing problems, with a focus on national security.

Scientist: Michael Navicky, navicky@hpc.msstate.edu or 662-325-0779
Responsible Unit: CCI
Project Information: Open-Source Intelligence Laboratory

COVID Emotions Project

As part of an NSF Rapid funded project #2031246, social and computer science research faculty and students at MSU developed a comprehensive database that captured 14,970,419 posts about emotions expressed (e.g., anxiety, sadness, fear, hope) during the COVID-19 pandemic from 10 social media and forums platforms from January 2020 to April 2021. The database entitled, “COVID-19 Online Prevalence of Emotions in Institutions Database” (COPE-ID) was constructed using Application Programming Interfaces and Python code to collect data based on selected keywords from the following platforms: 1) Twitter, 2) YouTube, 3) Reddit, 4) Parler, 5) 4chan, 6) 8kun, 7) Gab, 8) Tumblr, 9) Flickr, and 10) Mastodon. COPE-ID allows researchers to identify how people coped during various phases of the pandemic and how emotions are connected to and influenced by various social institutions (e.g., work, religion, family). The team integrates social, computer science, and data science research methods to address research questions surrounding the topic areas of emotions, mental health, misinformation, countering misinformation, and reactions to public health intervention efforts (e.g., stay-at-home-orders, gathering bans, mask mandates, and bar closures). More information about the COPE-ID database and resulting products can be found at https://copeid.ssrc.msstate.edu/.

Scientists: Megan Stubbs-Richardson (PI); Sujan Anreddy (Co-PI); Ben Porter (Co-PI)
Responsible Unit: Data Science for Social Sciences Laboratory at the Social Science Research Center
Project Information: COPE-ID Database

Using Data-Driven Statistical Models to Estimate Latent Variables in Health, Human Development, and Food Security

Faculty in the Department of Agricultural Economics at Mississippi State are pioneering new statistical models for the measurement of complex latent variables. What are latent variables? An analogy with driving will help to explain. Some characteristics of the world are explicit and straightforward to measure. A driver, for example, can measure the speed of an automobile using the speedometer. Other characteristics of complex systems like a car are impossible to measure directly. There is no single instrument able to measure something like “driver satisfaction.” We measure driver satisfaction indirectly using proxies like repeat purchases to form a picture. These latent characteristics are ways the Department of Agricultural Economics builds an understanding of vitally important issues like health, human development, and food security – qualities of the world that cannot be measured using a single outcome. Instead, these latent variables must be estimated by combining information from multiple observable variables to create a single estimate. Will Davis in the department of Agricultural Economics at Mississippi State University designs and applies Bayesian latent factor and item response theory models to measure latent traits at the individual, household, or population level. Using these models allows for the estimation of unobserved or latent traits within a data-driven framework, informing estimation using features of the data rather than subjective opinion about the underlying phenomena.

Scientist: Will Davis
Responsible Unit: Mississippi State University Department of Agricultural Economics
Project Information: Please contact Dr. Davis at gwd53@msstat.edu
Photo Credit: Caleb Stokes on Unsplash

Athlete Engineering

The mission of athlete engineering is to align researchers from all engineering disciplines, as well as kinesiology, sociology, and psychology researchers, with athletic practitioners such as strength and conditioning coaches, athletic trainers, and nutritionists, to enhance student-athlete safety and link human factors to performance technology. The Athlete Engineering team engages in data-science research in the areas of IoT sensor and wearable technology design and validation; unique methods for measuring ground-reaction forces and pressures; development of methodologies for testing wearable sensors and other human-performance-related solutions; and development of AI algorithms for motion analysis and classification.

Scientist: Reuben Burch
Responsible Unit: CAVS
Project Information: Athlete Engineering
Photo Credit: Megan Bean, Mississippi State University

Turbulence Modeling

You may have encountered turbulence when flying on a commercial airplane. The pilot announced that those random shudders and dips you experienced are due to air turbulence. The unsteady and violent movement of air or any fluid is known as “turbulence” and this phenomenon happens in environments as different as your own heart and arteries, rivers and streams, the flow of air over a wing, or the flow of water around a submarine propeller. Simulating the flow of material (computational fluid dynamics) is a classic use of high-performance computing as millions of dynamics calculations (force=mass x acceleration) are required to model the interactions of molecules of fluid or air as they encounter fixed or moving surfaces. Studying turbulence is a difficult endeavor – can we bring a propeller into a lab? How can we observe molecules colliding as they pump through the human heart? Yet we need to understand turbulence to create better, more efficient transportation and medical systems. That’s where the data science comes in. By representing the real world of molecules as data that can be examined in a mathematical context, data science can study turbulence down to the molecular level, find solutions, then translate these solutions back into the real world. Working with computational engineers at Mississippi State’s Center for Advanced Vehicular Systems (CAVS), Dr. Shanti Bhushan is developing advanced turbulence models for computational fluid dynamics (CFD) simulations. Models include the Algebraic Large Eddy Simulation (LES) model and a unique Dynamic Hybrid Reynolds-Averaged Navier–Stokes (RANS) / LES model that outperforms current LES and RANS models. In addition, Dr. Bhushan is performing innovative, basic research in bypass transition flow physics, one of the least understood problems in fluid mechanics, by using high-fidelity direct numerical simulation (DNS) studies.

Scientist: Shanti Bhushan
Responsible Unit: CAVS
Project Information: Turbulence Modeling

Estimating the Determinants of Telehealth Utilization

Defined as the provision of healthcare remotely by means of telecommunication technology, telehealth is a rapidly growing sector of the healthcare market. Telehealth illustrates the phenomenon of digital transformation in which even an industry historically characterized by the personal touch has been transformed using internet communications, mobile technology, and electronic health records. While one of the main roles of telehealth is increasing access to healthcare among individuals living in areas without sufficient access to in-person medical care, many of these individuals face barriers to telehealth adoption and participation. A major barrier is lacking access to the technology needed for telehealth, particularly for households without access to highspeed broadband internet. Will Davis in the department of Agricultural Economics at Mississippi State works to identify the roll that determinants like broadband access play in a patient’s decision to use telehealth services. This research relies on statistical models and other data science methods applied to novel data sets consisting of survey data, broadband access data, and local healthcare availability data. Current research funded by the University of Mississippi Medical Center and the National Institute of General Medical Sciences (NIGMS) focuses on improving our understanding of telehealth utilization decisions in the state of Mississippi. These results can be used to guide policymakers tasked with expanding broadband and telehealth access in the state.

Thermal Modeling for Additive Manufacturing

3d printing and other additive manufacturing techniques are revolutionizing the design and production of physical goods. One characteristic of digitally operated enterprises made possible by data science is that distinctions between virtual and real worlds become blurred. In some contexts, this means building digital twins of manufacturing systems to simulate new industrial methods. In the context of additive manufacturing research, this can mean simulating the use of materials to ensure the creation of precisely machined, high-quality objects that can operate in the same environments traditionally manufactured products can. The additive manufacturing (AM) group at Mississippi State’s Center for Advanced Vehicular Systems (CAVS) supplements its physical experimental capabilities with high-performance digital models to reduce the cost of trial-and-error physical experiments. The CAVS team performs temperature distribution calculations during Laser Engineering Net Shaping (LENSTM) deposition of titanium alloy-based components, aiming to find optimal ways to produce high-quality products. Using out-of-the box settings, these systems that use lasers to heat powdered metal to create solid metal objects can produce items with material defects, such as high porosity, insufficient strength, and high residual stress in the produced part. These defects can be minimized by using optimal settings discovered through digital modeling. CAVS’s approach is to implement numerical models of laser deposition, validate these models with data from physical experiments, and use simulations to find process parameters that will result in an optimal manufactured product.

Scientists: Additive Manufacturing Research Team
Responsible Unit: CAVS
Project Information: Additive Manufacturing

Biofluids Modeling

Computational Fluid Dynamics (CFD), the application of data science to model the interactions of molecules of fluid or air as they encounter fixed or moving surfaces, is a key part of studying and developing health therapies that involve bodily systems such as the lungs or the blood, environments in which biofluids like blood and saliva move under high pressure through channels in the human body. The COVID pandemic emphasized the importance of understanding the effects of forced respiration as patients, encumbered by the virus, faced additional risks from ventilators. Dr. Greg Burgreen and his team of computational engineers at Mississippi State’s Center for Advanced Vehicular Systems (CAVS) are using computational fluid dynamics (CFD) to simulate and improve novel medical therapies for both adults and children. This research has been instrumental in the awarding of several US and international patents on biomedical devices. Sub-projects include drug delivery in advanced models of diseased human lungs, cardiovascular artificial organ development, involving ventricular assist devices (artificial hearts) and oxygenation devices (artificial lungs), and state-of-the-art models of blood damage and thrombosis.

Scientist: Greg Burgreen
Responsible Unit: CAVS
Project Information: CFD Biofluids