Skip to main content

To encourage multidisciplinary collaborative research and to attract external funding to the university’s research, WPI launched the Transformative Research and Innovation, Accelerating Discovery (TRIAD) seed grant program in the fall. A semi-random selection process was used, a process that ensured that funded projects meet the administrative requirements of the program and guarantees portfolio diversity.  We have now funded 17 projects, all promising to be exciting innovative collaborations with faculty from all academic departments.

TRIAD Grants Will Expand Multidisciplinary Research Possibilities

Read this Herd story to find out more about how the TRIAD seed grant program started and how it helps advance WPI research.

A journey from high dimensional PDE

A Journey from High Dimensional PDE to Reinforcement Learning and Its Applications

Principal Investigator: Qingshuo Song

Collaborators: Yanhua LiSeyed A Zekavat

  • Read the description

    Reinforcement Learning (RL) has many Smart World emerging application. Examples include Autonomous Driving, Smart Structures, Smart Grids, urban computing, multi-agent network traffic control, and Robotics. These applications involve collecting data from a large number of sensors in order to optimally control devices or systems. Data volume collected in these applications is high, and real-time data analysis is vital to safe driving.

    This RL revolutionary technology surge can be amounted to the computational power in solving high dimensional Markovian Decision Process (MDP) using both the powerful parallel computing hardware and algorithmic improvements. As its counterpart of continuous time decision making, the stochastic control theory had been extensively studied in mathematical community and greatly enriched the literatures of in applied mathematics, partial differential equations and probability. It is well known that such a stochastic control problem leads to a very challenging area intertwined by fully nonlinear partial differential equation, namely Hamilton-Jacobi-Bellman equation (HJB), and nonlinear Feynman-Kac formula.

    Owing to the complex nature of optimal decision function, the solutions are often highly nonlinear and closed-form solutions are not usually obtainable even in the classical stochastic control problem. Thus, the numerical method by finite different/element method on its HJB has become a viable alternative. However, this method is usually not practical for high dimensional problems due to its well-known ``curse of dimensionality''.

    A recent trend to overcome this ``curse of dimensionality'' is to use deep neural network on its equivalent Backward Stochastic Differential (BSDE) formulation and proved efficient in a class of equations. It opens up the possibility to tackle high dimensional control problem using modern machine learning technology, but the stability and robustness becomes a widely open issue. Given that powerful solvability of reinforcement learning on MDP, a natural question is: Can one take this modern technology advantage in MDP to solve high dimensional stochastic control problem aforementioned?

    The answer is Yes in principle. Indeed, Markov chain approximation for computing stochastic control problem offers a systematic monotone scheme to approximate stochastic control problem by MDP. Later, this methodology and its convergence theory has been extensively studied to hybrid control problems. This idea opens up the utilization of recent developments in reinforcement learning theory, and provides feasible approach for high dimensional problems, which was not feasible ten years ago. Although the theory was established ten years ago, it still remains widely open for reinforcement learning computation of stochastic control in terms of stability, efficiency, and robustness. For instance, our experiment shows that MDP constructed from upward scheme effectively converges to stochastic control problem in a fixed finite time horizon (Parabolic HJB), but diverges for a random time horizon up to a stopping time (Elliptic HJB). To the best of our knowledge, no theory can support this phenomena. Therefore, our objective is to develop a feasible robust reinforcement learning MDP scheme by adapting Markov chain approximation for high dimensional stochastic control system.

Chemical Processing and Machine Learning alt
Chemical Processing and Machine Learning

Chemical Processing and Machine Learning

Principal Investigator: Randy C Paffenroth

Collaborators: Anthony G DixonMichael T Timko

  • Read the description

    The chemical process industry generates hundreds of billions of dollars each year in profit. At the same time, chemical processing contributes substantially to energy use, greenhouse gas emissions, and other forms of water, air, and soil pollution. Improved process technologies are needed to decrease the environmental footprint of chemical processing, while maintaining or reducing processing costs, and machine learning promises to be a transformative technology in this domain. Gas-liquid-solid reactions are an important class of reactions, important in the production of petrochemicals, pharmaceuticals, and specialty chemicals. Typically, gas-liquid-solid reactions involve reaction of a gaseous reactant with a liquid reactant on a solid catalyst. The catalyst is required to reduce operating temperature, and hence energy use, and improve selectivity, and hence decrease waste. While the chemical reactions themselves are important to reactor operation, many reactors suffer from poor transfer of heat out of (or into) the catalyst because of the complexity of the reactor system. Poor heat transfer leads to localized hot spots, where selectivity is sacrificed, catalysts can be damaged, and unsafe operating conditions can arise. The problem is especially challenging in gas-liquid-solid reactions since the flow field of both the liquid and the gas phase may be important to heat transfer rates. Accordingly, herein we propose to use machine learning to improve the performance of such systems.

    Current models and experimental investigations for heat transfer and reaction in fixed beds are based on the questionable use of effective parameters to lump together different heat transfer mechanisms. Such systems can be improved using machine learning and Data Science techniques to learn higher fidelity models then are currently practical. New, safer, and more efficient reactor designs would benefit from improved models of fixed bed heat transfer, arising from a better understanding of the interactions between the bed structure, flow fields and temperature profiles that arise from a data-oriented approach. The overall objective of this proposal is to advance fundamental understanding of heat transfer in fixed bed reactor tubes, taking into account the complexities of modern advanced catalyst particle shapes, such as multi-lobed particles or the presence of internal structure, using advanced machine learning techniques such as deep learning. The overall approach is to combine computational fluid dynamics (CFD) simulations of flow and heat transfer in fixed beds with machine learning methods, in particular artificial deep neural networks, to recognize fixed-bed flow patterns and connect them to the convective heat transport. The work will combine particle-resolved computational fluid dynamics (PRCFD) simulations of the multidimensional multiphase flows in tubes of complex particle shapes, with hybrid (grey box) modeling based on deep neural networks and first principles reaction engineering models.

Closed-Loop BCI Using Adaptive Kinetic Architectural Design to Regulate Human Emotional States alt
Closed-Loop BCI Using Adaptive Kinetic Architectural Design to Regulate Human Emotional States

Closed-Loop BCI Using Adaptive Kinetic Architectural Design to Regulate Human Emotional States

Principal Investigator: Ali Yousefi

Collaborators: Mohamad FarzinmoghadamErin Solovey

  • Read the description

    Emotion represents how we process information and how we interact with our surrounding environment. Emotion is linked to behavior and brain dynamics, and in our recent research, we have demonstrated how human emotional states can be characterized through neural activity and behavior (Yousefi et al., 2019). There has been a significant research effort to understand how environmental and habitual elements can impact human emotional states. Recent studies (Norwood et al., 2019; Coburn, Vartanian, & Chatterjee, 2017; Davies-Cooper et al., 2014; Joye, 2007) suggest that the built environment can significantly impact human psychology and cognitive behavior. For instance, Erkan (2018) showed the effects of the height of the space on wayfinding behavior. In another study, Coburn et al. (2019) found the relation between nature-like patterns and aesthetic preference. Shin et al. (2015) show the effect on emotions by altering direct or indirect light settings. In this research, we aim to systematically study the influences of architectural design elements, such as dimension, light, and color on human behavioral modalities and correlated neural activity. By understanding these mechanisms and characterizing possible causal dependence, we open doors for future research allowing us to utilize the architectural design features in a closed-loop brain-computer interface (BCI) to intervene in human emotional states and improve cognitive behavior (Shanechi, 2019). BCI aims to build a medium to communicate with the brain; if such a medium is being properly built, it can provide tools for many important scientific endeavors including promoting brain health (Widge et al., 2015; Shanechi, 2019).
    To conduct this experiment and record the data, we utilize new advances in the physiological signal recording - like EEG, heart rate, eye tracking - along with virtual reality (VR). Using VR, we are able to seamlessly - and inexpensively - change the environmental elements whilst simultaneously record from the brain and behavior. Using this recording platform, we collect necessary data to build preliminary (exploratory) models of the closed-loop BCI and more importantly to investigate connections between brain, behavior, and architectural elements.

Combating Online Health Misinformation:  An Integration of Behavioral and Computational Approaches alt
Combating Online Health Misinformation: An Integration of Behavioral and Computational Approaches

Combating Online Health Misinformation: An Integration of Behavioral and Computational Approaches

Principal Investigator: Nima Kordzadeh

Collaborators: Kyumin LeeBrenton Faber

  • Read the description

    People are increasingly searching for health information online. This information could be related to symptoms and consequences of diseases, medication alternatives and effectiveness, and general health and wellness topics such as those related to diet and nutrition. As early as 2008 researchers have documented people using the internet to compare, critically evaluate, and select their own medical treatments. Online health information is either presented by credible sources such as government agencies (e.g., U.S. Department of Health and Human Services), healthcare providers (e.g., Mayo Clinic), and independent healthcare websites (e.g., WebMD) or disseminated by non-experts through channels such as blogs and social media environments. In particular, social networking platforms such as Twitter, Instagram, and Facebook allow individuals to share health information, tips, and advice that are not necessarily truthful, reliable, and applicable in all contexts and situations. Misleading and inaccurate health information, also known as misinformation, are generally based on rumors, personal experiences, and unreliable sources. Using such unverified information could lead to negative health consequences for information consumers and the community as a whole. Our proposed project seeks to address this problem.

    Despite the growing use of online health (mis)information, research examining how people validate, trust, use, and share this information, particularly in social media environments, has been circular, inconclusive, or controversial.  Very few studies have empirically examined or verified these claims. Accordingly, the first objective of this research project is to fill this gap. To achieve this objective, we will draw on theories from information systems (e.g., IS Success Model and Source Credibility Theory) and cognitive sciences (e.g., Heuristic-Systematic Model of Information Processing) and conduct experimental studies to understand how people process and adopt online health (mis)information. Our proposed project will mainly focus on the health information shared via social media. The results, however, will enable us to develop an extended grant proposal and submit it to a national funding agency to study additional sources of health information such as healthcare websites and mobile applications. 

    Moreover, prior studies have rarely attempted to develop an intelligent information system involving computational mechanisms to actively direct online health seekers to credible and relevant sources of information and to proactively support healthcare advocates toward preventing continuous regeneration of health-related misinformation thus mitigating the impact of the misinformation. Some agencies such as the National Institute on Aging (NIA) and the Centers for Disease Control and Prevention provide general guidelines for health consumers to assess the trustworthiness of online health information. These guidelines, however, are not directly applicable in contexts like social media. Moreover, not all Internet users see these passive recommendations before searching for and using online health information. We believe a more effective mechanism should be in place to help individuals verify the accuracy, reliability, and completeness of the health information they see on the Internet and to enable them to find dependable sources of information. Accordingly, the second objective of this study is to (1) design and build a ranking algorithm, which searches and ranks relevant fact-checked pages or trustful information pages related to misinformation posted by healthcare seekers, and (2) build an effective recommender system, which proactively recommends related fact-checked information to healthcare advocates on social media who are interested in propagating correct health information.

    This research project has three knowledge pillars including 1) behavioral and information sciences, 2) computational and data sciences, and 3) medical and health sciences. Professor Nima Kordzadeh from the Business School with research expertise in behavioral and social aspects of information systems and health social media will lead the behavioral component of this project. Professor Kyumin Lee from the Computer Science department with research expertise in information retrieval, social computing and data science will lead the computational component of this project. Professor Brenton Faber from Biomedical Engineering and Humanities and Arts, with knowledge, research, and clinical experience in medicine and health sciences and a background in medical communication will lead the medical component of the project. He will evaluate the veracity of medical data and ensure that the research inputs, processes, and outputs are valid, useful, and conceptually consistent.

Complex Simulations for Local Environmental Policy: Digitization, Analysis, and Activism alt
Complex Simulations for Local Environmental Policy: Digitization, Analysis, and Activism

Complex Simulations for Local Environmental Policy: Digitization, Analysis, and Activism

Principal Investigator: Shamsnaz Virani Bhada

Collaborators: Constance ClarkJennifer deWinterPaul P Mathisen

  • Read the description

    This project brings together two forms of digitized information to analyze complex policy information as it intersects with green initiatives and provide data-driven recommendations for smart city development and growth.

    The first is Machine Readable Policy Models: Smart Connected Complex Systems such as health care or transportation are governed by a mosaic of policy documents that are Ambiguous, Vague, Complex, Verbose and Inconsistent. Inaccurate, Inconsistent and Unclear Policies lead to Systemic Failures such as Veteran Affairs Health Care Inconsistencies or failures in public transportation. Policies current and future need to be traceable, analyzable and digitized to reduce systemic failures for engineered systems. Policy Content Digitization and Analysis therefore provides a key touchstone in our research. We use systems approaches to build a Machine Readable Policy Models that are Analyzable, Traceable and Descriptive. The policy digitization framework represents the conceptual view supported by a step by step approach to achieve complete policy digitization and analysis.

    The second is simulation modeling of complex systems: Complex systems often emerge from the constraints imposed by economic, social, behavioral, technological, historical infrastructure, governmental, and historical exigencies. Policy decisions, then, are often created based on current usage data with little way to test shifts in the system within new contexts. Transportation, for example, is a key initiative in green cities and smart world infrastructure; however, the perception is that ridership is in decline and thus necessitates a reduction in service. Successful models in other locations are often dismissed, often for economic reasons, because the model of change is difficult to represent in new locations. Modeling and simulation uses mathematical and spatial representations of complex systems to test controllable variables in controlled systems.

    This project will bring together Machine Readable Policy Models and Simulation Modeling together through the test case of public transportation in Worcester. Worcester City Council has a Green Worcester plan in place, and public transportation is a key topic that is eluding action from policy makers. The team will work to analyze current interconnected policies, regulations, tax structures, available technologies. Following proposed smart connected systems, the team will build a city simulation model for public transportation to test possible solutions, with particular emphasis on race, gender, and socio-economic backgrounds, as well as key demographics around access to healthcare, education, work, and downtown 18-hour revitalization, collecting and analyzing data from different modeling. Our data analysis of the simulation will then inform the policy content modeling for policy recommendations put forward toward smart city planning and implementation.


Creating a Digital Platform for Understanding Fatigue in Lymphangioleiomyomatosis Patients

Principal Investigator: Ulkuhan Guler

Collaborators: Pratap M RaoBengisu Tulu

  • Read the description

    Tuberous sclerosis complex (TSC) is an autosomal dominant, multisystem disorder characterized by benign hamartomas in multiple organ systems, including the skin, brain, heart, kidneys, lungs and eyes. Incidence is estimated to be 1:6000. It is caused by mutations in tumor suppressor genes TSC1 or TSC2. Lymphangioleiomyomatosis (LAM), a multisystem disorder primarily affecting women, is characterized by cystic lung destruction, axial lymphatic abnormalities, and abdominal angiomyolipomas (AML). LAM occurs sporadically in patients with no evidence of germline genetic abnormality, and in about one-third of women with TSC [14-16]. Sporadic LAM is thought to be caused by a neoplastic cell in which TSC1 or TSC2 is mutated, more often TSC2 than TSC1.

    Fatigue is a common symptom in women with LAM. In fact, in one study 72% of women with LAM complained of fatigue. In this study, fatigue was reported by a high number of patients and was the most frequent symptom for older patients and those with TSC. Fatigue was reported by LAM patients of all age groups and, unlike dyspnea, was not strongly related to the degree of lung function decline, suggesting that fatigue is independent of lung dysfunction in LAM.

    Fatigue remains unexplained in LAM and TSC.  To date, no clinical study for subjects with LAM or TSC has particularly targeted the understanding of this very common manifestation that leads to a significant impact on the quality of life in patients with LAM. This constitutes an unmet need.

    We are proposing to develop a cost-effective miniaturized wearable device capable of wireless and continuous monitoring of vital parameters to optimize the outpatient management of LAM patients with fragile respiratory status.  The sensed data will be collected by a mobile device and transmitted to a secure server where the software will process the information to make it readable and digestible for doctors, who will then use that information to influence their treatment plan. The ability to measure accurate data wirelessly and noninvasively would allow a dramatic increase in the amount of data obtained, which will pave the way for new directions for medical studies.

    While the effects of fatigue on daily life have been described, very little is known about the effects of diet, exercise or other daily living factors on fatigue. We believe that understanding fatigue and how it relates to daily living factors, including but not limited to dietary choices and exercise habits, could pave the way to new therapies that could improve the quality of life in patients with LAM. However, lack of research tools to collect data about fatigue and daily living factors specifically relevant to TSC and LAM context prevents us from developing this important evidence base. Given the prominent use of mobile phones and mobile applications in the society, developing an application geared towards the needs of patients with LAM and TSC can result in approaching fatigue studies differently and can help researchers build a dataset to investigate the new hypothesis.


Detection of Bioelectrical Changes as a Function of Neoplastic Progression

Principal Investigator: Ahmet C Sabuncu

Collaborators: Izabela R StroeCagdas D Onal, Masqsood Ali Mughal

  • Read the description

    Early detection is the hallmark of successful cancer treatment. However, many tumors remain clinically occult until they are far advanced. Carcinoma is one of the most common types of cancer that proliferates in the epithelial tissues, with colorectal carcinoma cancer (CRC) being the 3rd most common cancer in males and 2nd in females, and globally the 4th cause for cancer death worldwide. New research shows that the incidence of CRC is increasing among younger adults for reasons that are not yet known. It is predicted that by 2030, CRC incidence rates will increase by 90% for people aged 20 to 34 years and by 28% for people aged 35 to 49 years.
    Endoscopy is the gold standard for CRC detection and removing of carcinoma and its premalignant precursor, dysplasia. However, its accuracy is suboptimal with 41% of precancerous adenomas are missed during examination. This is mostly due to: (1) the difficulty of identification of carcinoma in patients with preexisting conditions (such as inflammatory bowel disease, or in patients who have been previously biopsied), and (2) the inherent limitations of the technique that does not allow a full visualization of early carcinoma in a background of dysplasia and hidden polyps especially the ones behind colonic folds and flexures.
    We propose to develop a novel diagnostic tool for endoscopy that improves the detection accuracy of colorectal cancer and therefore increases the survival rate of cancer patients. Our proposed endoscope will incorporate an impedance spectroscope that scans the bioelectrical properties of tissues, such as tissue capacitance and conductance. It will enable a real-time bioelectric fingerprint of the colon for early detection of colorectal cancer, while minimally impacting the standard colonoscopy procedure.
    In this pilot, we will: (1) investigate the changes in the bioelectrical parameters of normal, precancerous and cancerous cells and tissues, (2) develop computational models to analyze and decompose impedance spectra into biologically relevant parameters of tissues, namely, cell membrane capacitance, transmembrane and extracellular conductance, and cytoplasm conductance. Correlations will be sought between these parameters and the severity of carcinoma. Results will be useful in formulating a hypothesis for the new endoscope design.
    These preliminary results will be used for future proposals to NIH and similar funding agencies.

Does mechanical stress facilitate muscle-neuron integration? alt
Does mechanical stress facilitate muscle-neuron integration?

Does Mechanical Stress Facilitate Muscle-Neuron Integration?

Principal Investigator: Suzanne F Scarlata

Collaborators: Kristen BilliarSarah D Olson, George D. Pins

  • Read the description

    Replacement of damaged and diseased tissue is a promising method to treat a variety of diseases including cardiomyopathy. For any type of tissue replacement, multiple cell types must be grown together and adapt similar properties which are required for the cells to function optimally. Muscle cells undergo regular stretch-contraction cycles, and thus the neuronal cells that regulate this contractility must have elastic properties to insure that connections to muscle cells remain intact. However, details of how neuromuscular connections behave with mechanical stress are lacking. In this project, we will to study the behavior of these connections in neuronal and muscle cells with mechanical stretch. The goal is to develop cell treatment and methodology to bring us closer to tissue replacement therapies.

    Experimental, we will follow the components of cells, their properties and connectivity using fluorescence-based imaging methods. Co-cultured cells will be viewed in real time using a mechanical stretch device. We would like t develop analytical models that will enable us to understand the interplay between the physical forces and the cell connections.

Environmental Stressors and Decision-making Performance

Environmental Stressors and Decision-making Performance in the Context of Climate Change

Principal Investigator: Shichao Liu

Collaborators: Soussan DjamasbiGbetonmasse Blaise Somasse, Sarah Strauss

  • Read the description

    The changing climate poses increasing environmental stresses on students and workers. Extreme heat events are projected to be more frequent in the future . Adverse environmental conditions such as high air temperature could impair decision-making and cause substantial labor productivity loss that is estimated to be worth 80 million jobs. The related cognitive impairment and the loss in productivity may have implications in terms of socioeconomic outcomes and social justice.

    The proposed project aims to understand the complex interaction between the physical environment and people’s group decision-making. We seek to improve the state-of-the-art modeling of individual and group decision-making by incorporating environmental factors such as temperature, humidity, lighting, acoustics, and air quality.

    The research team will apply multidisciplinary approaches to disentangle the complexity of the research topic. We will recruit students from WPI as human subjects and analyze their decision-making process in a laboratory lab under different environmental conditions. Their brain activities, physiological conditions, and emotional states will be monitored through bio-signal sensing including electroencephalogram (EEG), eye-tracking, electrocardiography (ECG), galvanic skin response (GSR), skin temperature, facial expression; these biophysical markers will be paired with qualitative interviews regarding subject perceptions of environmental conditions, to determine the extent to which awareness or concern for favorable or adverse environmental conditions might relate to performance outcomes.

    The research will help us to understand how the changing environment affects the performance and well-being of workers in the future. The proposed seed-grant project will support two Ph.D. and 2-3 undergraduate students and enable us to obtain preliminary results that would be used to apply for an external research grant from NSF or NIH.

How do plants sense pathogens

How Do Plants Sense Pathogens? Molecular Receptors and Cellular Signaling


Principal Investigator: Luis Vidali

Collaborators: Dirk AlbrechtSamuel Cabot Walcott

  • Read the description

    Over millions of years, plants have coexisted and coevolved with microorganisms. To survive and adapt, all plants have to be able to identify microbes and determine if they are pathogenic, beneficial, or neutral. The mechanisms for detection are complex and can be neutralized by pathogenic organisms. A highly-conserved mechanism that plants use to detect fungi comprises membrane receptors to fungal cell wall components; the most investigated of these is the receptor for chitin oligosaccharides. When the chitin receptor is activated, the signal is transduced by a rapid increase in cytosolic calcium --by a process similar to what takes place in neuronal and muscle cells. Downstream from this calcium increase, the signaling process results in changes in gene expression, including defense-related genes. The chitin response is conserved in all land plants, including the powerful model moss, Physcomitrella patens. We have generated transgenic P. patens plants expressing the cytosolic calcium fluorescent sensor, GCaMpf6. We found that soon after chitin treatment, the plants stop growing and produce oscillatory calcium spiking. We are currently analyzing the significance of these oscillations by evaluating the hypothesis that their frequency and amplitude encode information that is interpreted by downstream mechanisms. To fully develop this project, we need to assemble an interdisciplinary team with expertise in advanced three-dimensional microscopy, signal/noise processing, and mathematical modeling. It is anticipated that this project will provide sufficient preliminary results to submit a collaborative grant proposal to the NSF/USDA program on Plant Biotic Interactions.

ICME approach to design Novel safe Lithium electrode alt
ICME approach to design Novel safe Lithium electrode

ICME Approach to Design Novel Safe Lithium Electrode

Principal Investigator: James L Urban

Collaborators: Yu ZhongAlbert SimeoniMilosh T Puchovsky, Winston Soboyejo

  • Read the description

    Designing the novel electrode with increased voltage is the most effective way to improve lithium battery performance since the stored energy is proportional to the voltage, and the power is proportional to the square of the voltage. However, high voltage leads to the safety concerns of batteries. A cathode material’s safety is typically assessed as its resistance to releasing oxygen at elevated temperatures in its charged state. The oxygen released combusts the organic electrolyte and eventually leads to the thermal runaway of the cell to cause a fire. It has been identified that the cathode high oxygen partial pressure (PO2) or low oxygen release temperature will make the battery less safe. It is imperative to ensure battery safety by designing the novel cathode materials for lithium batteries.
    The long-term goal is to design a novel cathode material for the lithium battery with both high voltage and safety. To do that, we need to divide the research into two stages. Stage I: with the WPI internal support, we will figure out the mechanism of thermal runaway and define the role of thermodynamics in thermal runaway. Stage II: with the external funding from NSF/DOE to design a new cathode material with good performance and high safety with the ICME approach.

Multi-scale mechanical characterization and modeling of the brain for injury prevention alt
Multi-scale mechanical characterization and modeling of the brain for injury prevention

Multi-scale Mechanical Characterization and Modeling of the Brain for Injury Prevention

Principal Investigator: Songbai Ji

Collaborators: Yuxiang LiuYihao Zheng

  • Read the description

    Computer models of the human head are increasingly used to study the biomechanics of traumatic brain injury in order to better detect and prevent injury. Despite decades of efforts, however, head injury models are still far from sufficiently biofidelic. There are two primary roadblocks: (1) the brain material properties remain poorly characterized, even though they are crucial to the simulated brain mechanical behaviors; and (2) high-quality experimental data suitable for head injury model validation is lacking. To resolve these two roadblocks, we have formed a collaborative team, including experimentalists, Dr. Yuxiang Liu and Dr. Yihao Zheng from ME, and computer modeling expert, Dr. Songbai Ji from BME. We propose to characterize brain tissue properties at micro-scale using a unique technique based on optical fiber tweezers at varying locations (Liu) as well as at tissue-level using advanced tissue simulant and animal brains at injury-causing loading conditions (Zheng). These experimental data will then be integrated into a corresponding head injury model to enhance its biofidelity (Ji), which will be verified using data at the organ level (Zheng). This joint effort is expected to produce critical experimental data on brain mechanical behaviors at a range of spatial and temporal scales to enhance the head injury model biofidelity. Ultimately, this will improve brain injury detection and prevention in humans, as well as enhance the design of head protection gears.


MXenes: New 2D Materials for Applications in Catalysis, Sensing and Electronics

Principal Investigator: Lyubov Titova

Collaborators: Douglas T Petkie, Andrew R Teixeira, Aaron Deskins

  • Read the description

    MXenes is an emergent class of the two-dimensional transition metal carbides and nitrides with the general formula Mn+1XnTz (n = 1, 2, or 3; M=transition metal, X = C and/or N; Tz = surface termination, e.g., –OH, –F, =O). Only eight years since the discovery of the first MXene, Ti3C2Tz, the MXene family has grown to over 30 compounds.1 High metallic-like conductivity, hydrophilicity, flexibility and extremely high charge capacitance render MXenes attractive for applications such as supercapacitors, transparent flexible conductors and electromagnetic interference shielding.2-4 In addition, large surface area of individual nanosheets, presence of active terminations that can act as active sites for gas adsorption and nanosheet edges that can act as catalytic active sites makes these new materials promising for catalytic and gas sensing applications.5-9 Specifically, it has been recently demonstrated that Ti3C2Tz MXene nanosheets can catalyze electrohydrogenation of N2 to NH3 at ambient conditions.7-8 In a separate report, another MXene material Zr2CO2 demonstrated highly reversible release and capture of NH3, making it a potential NH3 sensor.10 In this project, we will work with the group of M. W. Barsoum at Drexel University and explore applications of MXenes as an electrocatalyst for energy-efficient synthesis of ammonia and other nitrogen compounds, as a sensor for NH3 and other gases, and as a materials for conductive electrodes and detectors. M.W. Barsoum is one of the people credited with discovery of MXenes in 2011, and his group will provide materials for this work. The results of this work will constitute preliminary data for collaborative proposals to DOE and NSF.

Photoacoustic Imaging of Neurovascular Bundles for Robot-Assisted Laparoscopic Pelvic Surgery alt
Photoacoustic Imaging of Neurovascular Bundles for Robot-Assisted Laparoscopic Pelvic Surgery

Photoacoustic Imaging of Neurovascular Bundles for Robot-Assisted Laparoscopic Pelvic Surgery

Principal Investigator: Haichong Zhang

Collaborators: Loris FicheraGregory S Fischer, Benjamin Nephew

  • Read the description

    The objective of this proposal is to explore the creation of a novel spectroscopic photoacoustic (PA) imaging solution to detect neurovascular bundles (NVBs) during robot-assisted pelvic surgical procedures including radical prostatectomies. This research is motivated by the high rate of nerve damage associated post-operative complications from pelvic surgery and the need for a reliable guidance system to assist surgeons in detecting and preserving these delicate structures. Especially in the case of prostatectomy, accidental damage to these delicate structures is a leading cause of post-operative erectile dysfunction in prostate cancer patients. There is a strong demand in the medical community for a technology to intraoperatively detect the location of NVBs. These structures are embedded in soft tissue (i.e. the “fascia” layers surrounding the prostate), and are thus invisible from the surface, making their detection challenging. Recent advances in medical imaging technology, namely the ability to harness the PA effect to enable high optical contrast with ultrasonic spatial resolution in deep tissue, opens new possibilities for the visualization of NVBs. The specific objective of the proposed one-year effort is to provide proof of concept evidence for the spectroscopic PA detection of NVBs that will support the systematic integration of this technology in robot-assisted laparoscopic nerve-sparing pelvic surgery such as radical prostatectomy. To evaluate the feasibility of detecting NVBs via PA imaging, we initially propose to spectroscopically characterize types of biological tissue commonly encountered during prostate cancer surgery and will then investigate methods to separate the PA contrast of the NVB from that of surrounding tissue. Our research is driven by the following two hypotheses: 1) The PA contrast of nerves and blood vessels, i.e. the main constituents of NVBs, can be selectively decomposed from that of surrounding structures; 2) Nerves, blood vessels, and other types of soft tissue surrounding the prostate each present a unique absorption spectrum in the Near-Infrared (NIR) range of wavelengths (700-1400 nm). While the ultimate goal of our research is to develop a PA imaging instrument that can be integrated in the workflow of robot-assisted pelvic surgery, this proposal focuses on verifying the above-mentioned hypotheses. It will culminate in a proof-of-concept of PA-based NVB detection in an in vivo rat model.

SUPER BUGS, Genomics-driven molecular and epidemiological modeling of multidrug resistance alt
SUPER BUGS, Genomics-driven molecular and epidemiological modeling of multidrug resistance

SUPER BUGS, Genomics-Driven Molecular and Epidemiological Modeling of Multidrug Resistance

Principal Investigator: Reeta Prusty Rao

Collaborators: Eric M YoungJian Zou

  • Read the description

    The key motivation for this proposal is to employ molecular and genomic approaches to understand mechanisms of multidrug resistance as well as manage, monitor and model the spread of Candida auris. C. auris has recently emerged as a drug resistant superbug in health care setting worldwide. The simultaneous emergence of multidrug resistant C. auris in multiple geographic locations is an epidemiological conundrum and a serious global public health concern (CDC, USA and PHE, Public Health England, UK report). The burden and threats posed by these superbugs has a significant impact on human health. Each year in the United States, an estimated 2 million people become infected with drug-resistant microbes resulting in at least 23,000 deaths. Our long-term goal is to leverage results of this study to understand the basis of drug resistance, discover safe, effective and resistance-resilient therapeutic agents and develop predictive models to monitor the spread of this superbug. More immediately, the preliminary data we collect will be used for an NIH R01 application.

    In response to this urgency, we initiated this collaborative TRIAD to integrate the expertise of Dr. Reeta Rao (Biology and Biotechnology) who will be focused on describing the molecular interactions at the host-pathogen interface, Dr. Eric Young (Chemical Engineering) who will use genomic tools to understand the basis of drug resistance and Dr. Jian Zou (Mathematics) who will develop statistical models to predict the onset and spread of C. auris related disease outbreaks.

    For our studies we will focus on two strains of Candida auris obtained from the Centers of Disease Control. Dr. Young’s laboratory will explore the genetic basis of C. auris drug resistance. We have a genomics pipeline that, in our analysis, returns superior genome assemblies than what currently exists. We will use this method to interrogate the genomes of two C. auris strains that each have different phenotypes to discover resistance-determining mutations. We will then perform genome edits to revert identified mutations to see if the drug resistance can be eliminated.

    Experiments performed in Dr. Rao’s lab will probe the interaction of the two C. auris strains with host immune cells. Our preliminary data indicates that the Caspofungin resistant strain, Cau3, is able to escape host phagocytosis as compared to the drug sensitive Cau1 strain. We will probe the interaction between -glucan a component of the fungal cell wall, and target of Caspofungin and Dectin 1, the pathogen recognition receptor on host immune cells.

    Dr. Zou’s lab will focus on modeling and predicting the onset and spread of C. auris related disease outbreaks. We will first examine uncertainty at varying levels of geographical granularity, and then developing approaches to accommodate the observed geospatial uncertainty into the methodology. This includes the ability to characterize other key outbreak dimensions such as peak, incubation period, duration and intensity of the outbreak. We will conduct sensitivity analyses regarding model and network mis-specification and provide tools to trace the effects of spatial and temporal uncertainties in data and model parameters.


Solid electrolyte with high ionic conductivity and stability through high throughput experiments, phase modeling and AI

Principal Investigator: Yan Wang

Collaborators: Yu ZhongXiangnan Kong

  • Read the description

    The rechargeable lithium ion battery market was approximately $11.8 billion in 2010 and is expected to grow to $53.7 billion in 2020. It is evident that lithium ion batteries will dominate the electric car industry within the next 10~20 years, due to their high energy density and as auto makers strive to meet CO2 emission standards. Although random explosions from overheating haven’t been a widespread problem, Sony recalled 7 million lithium ion batteries on laptops due to the safety concern in 2006. In the case of hybrid and electric vehicles, for example, the GM volt, Nissan Leaf, and Tesla Roadster, large lithium ion battery packs with hundreds or thousands of cells are used. Recently, Boeing 787 was pulled out of service due to the fire of its lithium ion batteries. Therefore, safety with the use of lithium ion batteries in transportation industry is a big concern.
    Solid state batteries are intrinsically safe and offer high energy density due to the usage of lithium metal as the anode. Currently, ceramic solid electrolyte is widely developed. However, ceramic solid electrolyte is the bottleneck for developing high-performance solid-state lithium ion batteries due to the low ionic conductivity, instability and interfacial resistance, which prevents them from being commercially adopted.
    New material manufacturing from discovery to marketable product normally needs 15-20 years. Material synthesis and testing could take much time during the process. For example, conventional solid electrolyte materials are normally synthesized by sintering, which could take a few days. After solid electrolyte materials are synthesized, they will be tested and characterized, which could take weeks or months, even years. Therefore, fast material synthesis, and testing combining with phase modeling and artificial intelligence (AI) prediction are needed to predict and optimize the next generation solid electrolyte materials.
    The goal of this project will be expedited the discovery of new solid electrolyte materials. The team will synthesize different solid electrolyte through developed high throughput sintering methods guided by phase modeling. Then the synthesized electrolyte will be tested to determine its ionic conductivity and stability. Finally, the generated electrochemical data will be used to screen, determine and predict the optimized cathode materials with developed AI techniques.

Zero- or Negative-Emissions Aerospace Fuel

Zero- or Negative-Emissions Aerospace Fuel

Principal Investigator: Jagannath Jayachandran

Collaborators: Adam C PowellRonald Grimm

  • Read the description

    MgH₂ has relatively high energy density (22% higher than jet fuel per unit volume, 32% lower per unit mass), and when burned, produces MgO particles. A mixture/slurry with jet fuel should be pumpable in tanks and stable in air, where bare MgH₂ powder would otherwise spontaneously oxidize and pose a major dust explosion hazard. During combustion, MgH₂ decomposes at around 600 K, creating hydrogen which will potentially burst and fragment the magnesium particles. The remaining fine magnesium particles melt at 923 K then evaporate at 1365 K (or ~1900K at the 20 bar engine pressure, though boiling point will be reduced by small size), so combustion should occur in the vapor phase and produce extremely fine MgO particles. It is expected (from thermodynamic analysis) that the MgO particles will react with CO₂ in the plume and during settling process, before depositing onto the earth’s surface. If the fuel is 55% MgH₂ by mass (~45% by volume), it will achieve net zero emissions; if more, then net negative emissions. MgH₂ can be produced from sea water using electricity with zero direct emissions. The oceans have two quadrillion tonnes of Mg, i.e. 300,000 tonnes/human, so this is a limitless resource.

    This TRIAD project will employ 1.5 graduate students over one year to perform atomization and combustion experiments, techno-economic modeling, and atmospheric chemistry modeling to begin to answer multiple research questions in order to put together a strong proposal to DOE, DOD, or NSF. The objectives of the current TRIAD study are summarized below:
    • Atomization: A device to atomize this high-viscosity low-surface tension slurry will be identified, procured or built, and tested. Particles are expected to interact strongly in the strong shear flow, such that a small nozzle could clog. An ultrasonic or electro-spray mechanism will likely be required.
    • Combustion: Experiments will be performed to investigate the combustibility of the MgH₂ mixed with hydrocarbon slurry. Specifically, laminar spray flames of the mixtures of atomized slurry (small droplets) and air will be established in order to determine the laminar flame speed (SL), which is a measure of the exothermicity, reactivity, and diffusivity of the combustible mixture.
    • Production: Existing literature will used to develop a flow sheet for a production of MgH₂ from sea water using electricity with no accompanying direct emissions. The economic and energy consumption aspects of the production process will be evaluated and new lower (energy/economic) cost alternative processes will be explored.
    • Environmental impact: Experimental studies of the reaction of gas-phase CO₂ and particulate MgO will be studied under a range of temperatures. X-ray photoelectron spectroscopy will be used to quantify the products of reaction and correlate the quantity of carbonate carbon to magnesium for the products of reaction at varying time and temperature. The outcome of these studies will enable a deeper understanding of a chemical transformation relevant to the interaction of the exhaust with the atmosphere.