DOI
https://doi.org/10.25772/2GV5-Q687
Author ORCID Identifier
https://orcid.org/0000-0002-6342-4850
Defense Date
2024
Document Type
Dissertation
Degree Name
Doctor of Philosophy
Department
Chemical and Life Science Engineering
First Advisor
Thomas D. Roper
Second Advisor
Laleh Golshahi
Third Advisor
Adil Mohammad
Fourth Advisor
Ram B. Gupta
Fifth Advisor
Frank Gupton
Abstract
My PhD studies have been one of the most transformative periods of my life. Traditionally, earning a PhD requires the creation of new knowledge and a substantial contribution to the field. While the knowledge generated is undoubtedly valuable, the deeper purpose often lies in the personal growth achieved through this process. By personal growth, I refer to the development of the methods and discipline that are necessary for continuing to generate new scientific knowledge throughout one's lifetime. In this dissertation, I aim to demonstrate that I have not only significantly contributed to the field of chemical and life science engineering but also that I have done so across multiple subdisciplines. These contributions have been made possible through my personal growth, the outstanding mentorship I received, and the successful funding of my own research.
The transition of topics between some chapters reflects my evolving interests and skillsets as I progressed throughout the PhD program. The first section of my dissertation (chapters 1-3) is focused on the development of pharmaceutical manufacturing processes and the analytical technologies to monitor them. The first chapter “An Automated Continuous Synthesis and Isolation for the Scalable Production of Aryl Sulfonyl Chlorides” was my contribution to a larger project that focused on developing a continuous manufacturing process for a critical active pharmaceutical ingredient (API). For this project, the objective of employing continuous manufacturing was to help reduce manufacturing costs—which can help to bring pharmaceutical supply chains back into the United States. My primary contributions to science in this chapter were 1) the development of a process control scheme that utilized weighing balances to maintain CSTR levels, 2) improving the environmental metrics of the chlorosulfonation reaction by optimizing yield and quench conditions and 3) the scale-up of a historically hazardous reaction using flow chemistry. This chapter is adapted from its published form that originally appeared in Molecules, Importance of Flow Chemistry: Active Pharmaceutical Production.
The second year of my PhD program, I was awarded an Oak Ridge Fellowship for Advanced Pharmaceutical Manufacturing (ORISE) to spend a year at the Food and Drug Administration under the supervision of Dr. Adil Mohammad. During this year, I produced two additional peer-reviewed publications that were accepted in Reaction Chemistry and Engineering and Organic Process Research and Development. These two works are modified for chapter 2 and chapter 3, respectively. The focus of the project during my ORISE fellowship was to investigate the effects of continuous manufacturing on drug product quality for carbamazepine. In these chapters, I first developed validated high-performance liquid chromatography (HPLC) and Raman spectroscopy methodologies that served as the foundation for the later works. Using these analytical techniques, I developed a kinetic model for the batch and continuous syntheses. The kinetic model was applied to simulate system disturbances and evaluated with the two analytical methods. Following the development of a baseline synthesis procedure, several optimizations were applied to maximize the yield and purity of the undesired CBZ polymorph from the crude reaction mixture. Alongside Harrison Kraus, I designed and implemented a batch and continuous recrystallization procedure for the synthesized material to obtain the desired polymorph and purge the undesired impurities. My primary contributions to science in these chapters were 1) implementation of real time Raman spectroscopy for quantification of carbamazepine and the iminostilbene starting material, 2) a novel methodology for fitting kinetics to a reaction network where we were unable to detect the degradation products from one of the two pathways, and 3) continuous polymorphic recrystallization of material with impurities that was also synthesized using flow chemistry.
After my ORISE fellowship and my return to VCU, with the help of Dr. Roudabeh Moazeni-Pourasil, it was discovered that a methodology which I had envisioned for Raman model development for monitoring API production had novel applications towards the automation of chemometric model development. This discovery created several opportunities for the development and funding of my own research projects. This research has led to receiving a Virginia Innovation Partnership Corporation (VIPC) Commonwealth Commercialization Fund (CCF) $75K grant with $75K VCU matching and two smaller awards of $20K and $30K. Furthermore, I drafted a Food and Drug Administration Broad Agency Announcement (BAA) early concept paper that was recommended for full submission with the final decision pending ($1,600K) based on this technology. Lastly, I have generated two provisional patents, one of which has been converted to a full patent by the University.
Chapter 4 details work completed with the $20K award to analyze and apply machine learning to muscular dystrophy patient samples. Collaborators Dr. Melissa Hale and Dr. Nick Johnson sought ways to monitor the progression of muscular dystrophies through biofluids, which are less invasive than muscle biopsies. In this project, I proposed using surface-enhanced Raman spectroscopy (SERS) of human plasma, secured the funding, and led the research. The analytical measurements were conducted by Dr. Roudabeh Moazeni-Pourasil and I under the guidance of Dr. Massimo Bertino at the Nanomaterials Core Characterization (NCC) facility. This project required the approval of special protocols for analyzing biosafety level 2 (BSL2) samples at the NCC, which I played a lead role in developing. I consider this chapter one of my most significant contributions to science, as it entailed a successful, first-of-its-kind quantitative study for a large dataset of a rare disease using human samples. My success in designing and executing this project was significantly aided by my prior experience developing Raman models for API quality control.
Chapter 5 details the development and execution of my novel methodology— iterative regression of corrective baselines (IRCB)—which proved to be highly useful in automating chemometric model development. This work was funded through a VIPC CCF award that I applied for and received. This funding primarily focused on validating the intellectual property and developing a minimum viable product (MVP) for pharmaceutical data analysis. The MVP was a graphical user interface (GUI) to employ the technology, that I primarily developed with assistance from Hayden Nothacker in the later stages. In chapter 5, the IRCB is introduced and demonstrated using three regression case studies, one that was produced by myself and Dr. Roudabeh Moazeni- Pourasil and two which rely on previously published datasets. The objectives of IRCB are to 1) accelerate chemometric model development by serving as a highly robust and universal preprocessing method, 2) improve the compatibility of spectral data with non- linear machine learning models such as random forest and XGBoost (XGB), and 3) provide more insights into the data by projecting the baseline regions used to build the model back onto the original data. In this chapter, IRCB is proven to be a highly useful tool for the development of quantitative models from several types of spectroscopic data. It is demonstrated that IRCB can improve the compatibility of spectral data with non-linear machine learning algorithms, such as random forest.
Chapter 6 details additional use cases for the IRCB methodology. This chapter summarizes the current state of development for IRCB, providing a foundation for future graduate students to further investigate. In section 6.1, the performances of the key IRCB algorithms are benchmarked using a variety of high-performance computational approaches. The status of the graphical user interface (GUI) for IRCB regression model develop is described in section 6.2. Next, I propose future directions to explore using IRCB and report the preliminary work that has been done for each use case. These applications include outlier detection for regression models (6.3), classification of spectral data (6.4), regression modelling and mapping of three-dimensional spectra (6.5) and the analysis of DNA methylation data (6.6). I next apply IRCB with several machine learning strategies to the SERS muscular dystrophy dataset (6.7) that was reported in chapter 4.
I conclude (Chapter 7) by summarizing the primary contributions to science described within the first six chapters and then finally actionable, specific recommendations for the continuation and completion of these projects.
As I reflect on my PhD journey, I am filled with gratitude for the mentorship, support, and opportunities that have shaped my experience. This dissertation not only encapsulates the diverse and significant contributions I have made to the field of chemical and life science engineering but also serves as a testament to the personal growth I have undergone. The skills and knowledge I have acquired, along with the resilience and adaptability I have developed, will undoubtedly guide me in my future endeavors. I am excited to continue exploring new frontiers in science, driven by the same curiosity and passion that led me to pursue this PhD.
Rights
© The Author
Is Part Of
VCU University Archives
Is Part Of
VCU Theses and Dissertations
Date of Submission
10-19-2024