Research

(i) Cancer Therapeutics

Seed_grant_picture

Project Summary: In 2019, there will approximately be over 1.7 million new cancer cases diagnosed and 600,000 cancer-related deaths in the United States. These numbers are expected to rise consistently every year, which leads to higher demands for advancements in medicine. Existing treatment methods have displayed severe side effects that reduce the quality of life. Using nonlinear programming in MATLAB we have been trying to optimize a cancer treatment schedule for individual patients to keep leukocyte count high while reducing tumor size. Neural networks and decision trees were also used to perform a statistical analysis in a program called R to determine which biological markers are linked to Acute Myeloid Leukemia (AML) and Acute Lymphoblastic Leukemia (ALL). These two methods are forms of machine learning that can determine the likelihood of a biomarker being found in a patient with either of these two cancers by introducing a dataset from clinical studies. This information is necessary for determining which tests should be run to determine if a patient has these cancers and which tests are most effective for diagnosis.

(ii) Water-Energy Nexus

wastewater graphical abstract

Project Summary: Water is the most necessary commodity for survival for all forms of life around the world. Unfortunately, it is often undervalued and wasted, as those who readily have access to clean, safe water do not understand the magnitude of its’ importance. The pharmaceutical industry, for example, uses water in almost every manufacturing process, as a solvent for reactions as well as a cleaning agent for washing mechanical equipment. This water is often filled with ingredients that will prove detrimental to life forms waiting downstream from this wastewater. Conversely, one of the main challenges that many nations around the globe are facing is a lack of access to clean water. Thus, providing clean water through wastewater treatment (WWT) systems is a crucial step for continuing towards a sustainable future. Treating wastewater depends on the wastewater involved, which will determine the treatment technology to be used, as the contaminants present must be treated differently. Certain factors like toxicity, size, chemical structure, and physical amounts all play a role in how the problem is approached. Therefore, the development of a superstructure, consisting of all treatment methods and then reducing to the desired technologies in order to best select the route that can most effectively process the wastewater is essential. The technologies in consideration such as filtration units, advanced oxidation processes, biological methods, and many more comprise to a full system able to effectively treat both pharmaceutical and municipal wastewater. Through the use of material and energy balances, as well as cost and design equations, while still focusing on the importance of environmental and societal impacts; information that is derived from previous reports and data, we are able to develop a superstructure model to select the desired technologies for the type of wastewater present. We use representative case studies for municipal and pharmaceutical wastewater treatment and model them as optimization problems in the GAMS (General Algebraic Modeling Systems) programming language. In the next step, we use the P-graph approach for solving the same problem as this tool can provide insights into non-intuitive solutions, which will guarantee global optimality. P-graph approach can provide a ranked list of networks with detailed costs which currently cannot be obtained in GAMS.

(iii) Roadmap for Solvent Recovery in Industrial Manufacturing

Solvent Recovery Graphical Abstract

Project Summary: The global chemical market is projected to double between 2017 and 2030 but waste generation due to poor solvent selection and processing inefficiencies in the chemical industry have led to a growing concern for chemical releases, exposures, environmental impacts, and health safety. The US EPA has estimated that solvent emissions resulting from the chemical market growth can reach up to 10 million metric tons of carbon dioxide equivalent. As an alternative to conventional solvent disposal methods such as incineration, solvent recovery is being considered to improve the greenness and overall sustainability of processes in the pharmaceutical and fine chemical industries. A typical solvent recovery process requires multiple stages of separation, which may include several applicable technologies. A superstructure approach was used to develop the framework for solvent recovery, which begins with material (waste stream) input, followed by a number of process technologies that have the capability to perform the desired separation, reaction, or mixing. Based on the property of the stream that exits each technology, additional pathways are presented to reach the desired output (recovered solvents). Given that a superstructure contains many possible combinations of pathways, analyzing each pathway one-by-one is not a feasible approach. To assess the optimal solvent recovery path, mixed-Integer non-linear programming (MINLP) problems were formulated around solvent recovery case-study specific superstructures and solved through the General Algebraic Modeling Systems (GAMS). The systematic framework developed in this study can be applied to existing and future chemical process designs to minimize solvent waste, process cost, and environmental impacts.

(iv) Optimizing Resin Selection for Ion Exchange Processes

ion exchange resin abstract

Project Summary: Ion exchange resins are widely used in the industry to purify water and recover some valuable compounds. Most of these resins are polymeric based beads that are characterized by various parameters such as particle size, pore diameter, functional groups, and polymeric matrices that play an important role in a successful purification process. Within the last 100 years, many process models have been identified to relate certain resin characteristics to the process performance. However, there have been little to no significant work done on designing or characterizing the resin needed for desired outcomes. This research project will concentrate on identifying working models that identify such direct relationship between process performance and the key resin parameters.

(v) Management of IBS (Irritable Bowel Syndrome) using predictive analysis for gut microbiome fingerprints and probiotic supplements

Management of IBS (irritable Bowel Syndrome) using predictive analysis

Project Summary: IBS is a chronic disease which has a multifactorial etiology, correlating these factors to understand the exact cause and determine the most suitable treatment or patient management strategy is of high importance. Systems engineering tools can be of help in the systematic analysis of IBS symptoms, cause-effect and treatment responsiveness in patients.

This work is in collaboration with the Cooper Medical School at Rowan University (CMSRU) and has been funded by the Inspira Health Foundation. IBS patient data as well as healthy control samples are collected by our collaborators at CMSRU and tests are performed to determine the bacterial composition of the gut microbiota. Subsequently, data analysis is performed using machine learning algorithms such as Nearest Neighbor Methods, K-means clustering, Artificial Neural Networks, Naïve-Bayes estimation, etc. These help in identifying the important correlations between the different patient-specific characteristics and the IBS disease characteristics.

Model development will involve the use of these correlations as well as the effects of proposed treatment therapies on the patient. These will be in the form of mathematical equations which will be time-dependent, and we will be able to visualize the course of the disease in presence as well as the absence of a treatment strategy. The desired well-being of the patient will be represented in some of the model variables and the range of these values will determine the overall patient health.

Based on this model, we will apply optimization methods which would help in predicting the best treatment method. Additionally, we will perform sensitivity analysis to test the accuracy of the proposed model and optimization strategy.