UC Irvine

The coronavirus disease 2019 (COVID-19) pandemic caused millions of deaths and serious economic disruptions, also boosted the unprecedented development of novel nucleic acid detection methods. Although polymerase chain reaction (PCR) is the golden-standard and has been widely used for practical nucleic acid detection, isothermal amplification strategies capable of rapid, inexpensive, and accurate nucleic acid detection also provide new options for large-scale pathogen detection, disease diagnosis, and genotyping. Here we are going to describe an assay development journey from a simple COVID-19 detection assay to a genotyping strategy and eventually to a droplet-based amplification-free assay.

First, we described a highly sensitive multicomponent XNA-based nucleic acid detection platform that combines analyte preamplification with X10–23 mediated catalysis to detect the viral pathogen responsible for COVID-19. It is termed RNA-Encoded Viral Nucleic Acid Analyte Reporter (REVEALR), functions with a detection limit of ≤20 aM (∼10 copies/μL) using conventional fluorescence and paper-based lateral flow readout modalities. With a total assay time of 1 h, REVEALR provides a convenient nucleic acid alternative to equivalent CRISPR-based approaches, which have become popular methods for SARS-CoV-2 detection. The assay shows no cross-reactivity for other in vitro transcribed respiratory viral RNAs and functions with perfect accuracy against COVID-19 patient-derived clinical samples.

Second, we explained how we design REVEALR into a novel genotyping assay that detects single-base mismatches corresponding to each of the major SARS-CoV-2 strains found in the United States. Of 34 sequence-verified patient samples collected in early, mid, and late 2021 at the UCI Medical Center in Orange, California, REVEALR accurately identified the correct variant. The assay, which is programmable and amenable to multiplexing, offers an important new approach to personalized diagnostics.

Third, we talked about an improved REVEALR platform, termed digital droplet REVEALR (ddREVEALR), that can achieve direct viral detection and absolute quantitation utilizing a signal amplification strategy that relies on DNAzyme multiplexing and volume compression. Using an AI-assisted image-based readout, ddREVEALR was found to achieve 95% positive predictive agreement from a set of 20 nasopharyngeal swabs collected at UCI Medical Center in Orange, California. We suggest that the combination of amplification-free and protein-free analysis makes ddREVEALR a promising approach for direct viral RNA detection of clinical samples.

Finally, we summarized the developed DNAzyme-based nucleic acid detection methods, offered some alternatives, compared the DNAzyme-based platforms with CRISPR based platforms, and gave insight on potential future directions to further elevate the REVEALR system.

This dissertation investigated second graders’ oral discourse production regarding three aspects: 1) oral discourse production by discourse context, 2) dimensionality of oral discourse production, and 3) contributions of language and cognitive skills to oral discourse production. Data came from 330 second grade students (53% boys) from the Southeastern part of the United States, and oral discourse production was measured through picture description task which was transcribed verbatim and coded for linguistic (e.g., adverb, conjunction, pronominal, elaborated noun phrase, mental state talk) and discourse features (e.g., proper character introduction, degree of decontextualization, perspective taking). Study 1 examined how second graders vary linguistic and discourse features depending on discourse conditions: contextualized condition—describing the picture to an examiner while looking at it together—and decontextualized condition—pretending to describe the picture to a friend while sitting in front of the examiner. Results of multilevel regression showed that type-token ratio, higher degrees of decontextualization, and complex perspective taking were higher in the contextualized condition whereas some elaborated noun phrases, coordinating conjunctions, non-clauses, and proper character introduction occurred more frequently in the decontextualized condition, controlling for total productivity and student demographics. The findings illustrated the extent to which children used their discourse knowledge in oral discourse production. Study 2 looked at the factor structure of oral discourse production using a total of nine linguistic and discourse features identified from the same picture description task in Study 1. Results from confirmatory factor analysis showed that a bi-factor structure consisting of a general oral discourse production factor and two specific linguistic features and discourse features factors provided the best fit to the data. The only reliable factor was the general factor reflecting the common variance among the linguistic and discourse features. Study 3 investigated the structural relations of children’s domain-general cognitive skills (working memory, attentional control), foundational language skills (vocabulary, grammatical knowledge), and higher-order cognitive skills (knowledge-based inference, perspective taking, and comprehension monitoring) to oral discourse production. Oral discourse production was measured by the same picture description task as in Study 1. The results from structural equation models showed that domain-general cognitive skills and foundational language skills had indirect contributions to oral discourse production through the higher-order cognitive skill of inference, highlighting the supporting role of language and cognitive skills and their direct and indirect relations to oral discourse production.

The #MeToo movement, an online social movement related to sexual violence, went viral at a time when several high-powered men were accused of sexual assault and harassment. There were significant tensions within and about the movement that played out in the media. This project employed content media analysis to explore the tensions, particularly around punishment, healing, due process, and counternarratives. The analysis revealed that while accountability was important in the movement and in the discourse, other kinds of justice were also necessary, especially the need to help and protect others. Due process for men accused of sexual violence was a major theme identified in this study. Though the potential for unfairly punishing someone without cause or evidence is a legitimate concern, the discourse and backlash tended to use this concern as a way to shut down the conversation, to suggest that sexual violence was not as a big a problem, and to obscure the goals of leadership and many of the survivors—to support each other and heal. Also obscuring these goals was the backlash of the movement. Analyzing these key themes led me to develop two main concepts. First, survivor justice an abolitionist perspective focused on healing, survivors helping other survivors, and community accountability. Second, I argued that the due process claims reflect a kind of legal consciousness of legal procedures applied to explain everyday interactions, I call this social due process.

This research investigated the gas diffusion layer (GDL) membrane intrusion into a gas flow channel (GFC) using physics-informed machine learning for PEM fuel cells. This study was done to reduce the time it takes to simulate the model and find the intrusion area. To establish the training data for machine learning, different configurations of the GFC and GDL were created using COMSOL to simulate the GDL intrusion into a gas flow channel. The data from these simulations then get exported into Design-Expert, a statistical software to determine the parameters with the highest impact on the % intrusion. The GDL's Young's Modulus (E_GDL) was found to be the parameter with the most significant impact, with an F-value of 3776.66 compared to the next parameter Bipolar Plate Channel Width (CW_BP) with an F-value of 992.59, or the lowest parameter GDL Height (H_GDL) with an F-value of 159.96. Design-Expert also shows that (E_GDL) and the Bipolar Plate Rib R_BP positively affect the \% intrusion area, where both will yield lower % intrusion the higher the parameters are. Whereas higher the values are (CW_BP), (H_GDL) and Pressure (P) will increase the % intrusion.

With the finding from the Two-Factorial test, the simulation interval for (E_GDL) is higher than the other parameters to yield more accurate training data for machine learning. Then performing, a parametric study will be done to find the x-y coordinates of the intrusion curve to export to MATLAB to find the intrusion area. Four machine-learning algorithms, linear regression, Decision Tree, SVR, and KNN, were deployed to train using 70% of the data set and the remaining 30% for testing. The accuracy of each model were calculated based on how close the prediction is compared to the actual value. It was found that out of the four algorithms, Linear Regression has the lowest model accuracy at 68.5% and the highest RMSE at 0.0751, and Decision Tree has the highest model accuracy at 95.5% and lowest RMSE at 0.0303. Thus, Decision Tree was used to make predictions for various ranges of the five parameters to find the optimization parameters for design. This study was done to help reduce the time it takes to simulate the model and find the intrusion area. It takes an average of 20 minutes to create the model, simulate it, and calculate to find the intrusion. With the trained machine learning models, the intrusion area can be found in less than a minute. The machine learning model also identifies the parameter ranges for less than 10% and 20% intrusion to guide fuel cell material selection and design.

Reinforcement learning (RL) has proven useful for a wide variety of important applications, including robotics, autonomous vehicles, healthcare, finance, gaming, recommendation systems, and advertising, among many others. In general, RL involves training an agent to make decisions based on a reward signal. One of the major challenges in the field is the sparse reward problem, which occurs when the agent receives rewards only occasionally during the training process. This can make conventional RL algorithms difficult to train since the agent does not receive enough feedback to learn the optimal policy. Model-based planning is one potential solution to the sparse reward problem since it enables an agent to simulate their actions and predict the outcome far into the future. However, planning can be computationally expensive or even intractable when too many time steps are required to be internally simulated, due to combinatorial explosion.

To address these challenges, this thesis presents a new RL algorithm that uses a hierarchy of model-based (manager) and model-free (worker) policies to take advantage of the unique advantages of both. The worker takes guidance from the manager in the form of a goal or selected policy. The worker is computationally efficient and can respond to changes or uncertainty in the environment to carry out its task. From the manager’s perspective, this abstracts away the trivially small state transitions, reducing the depth needed for tree search, and greatly improving the efficiency of planning.

Two different applications were used for evaluation of the hierarchical agent. The first is a maze navigation environment, with continuous-state dynamics and unique episodes. This makes the environment extremely challenging for both model-based and model-free algorithms. The performance of the agent was evaluated on multiple platforms for the random maze task, including DeepMind Lab. For the second demonstration, the proposed algorithm was compared against other algorithms with the Arcade Learning Environment, which is a popular RL benchmark. In comparison with state-of-the-art algorithms, the proposed hierarchical algorithm is shown to have a faster convergence and greater sample efficiency on several tasks. Overall, the proposed hierarchical approach is a potential solution to the sparse rewards problem, and may enable RL algorithms to be applied to a wider range of tasks, ultimately leading to better outcomes in various applications.

There are numerous applications for deep learning in a healthcare setting including: providing more accurate diagnoses, recommending treatment plans, predicting patient outcomes, tracking patient engagement and adherence, and reducing the burden of administrative tasks. This plethora of applications has resulted in the widespread publication of deep learning algorithms applied to healthcare data. Despite numerous publications showing deep learning to be very successful in retrospective healthcare studies, very few of these algorithms are then actually incorporated into clinical practice. While there are many factors influencing the lack of algorithm deployment, one of the major reasons is a lack of trust in deep learning. This lack of trust stems in part from a lack of model interpretability and an inability to independently verify published results due to a lack of data availability. In this work, we explore generalized additive models with neural networks (GAM-NNs) as a method of improving model interpretability and we propose MOVER: Medical Informatics Operating Room Vi- tals and Events Repository a publicly available repository of medical data designed to improve visibility into deep learning algorithms in healthcare.

Similarly, deep learning can be used to analyze circadian omic (e.g. transcriptomic, metabolomic, proteomic) time series data. Several studies have shown that a disruption to circadian rhythms have been linked to health problems such as cancer, diabetes, obesity, and premature aging. In order to gain clinician trust in the conclusions drawn from circadian omic analyses we propose CircadiOmics: the largest annotated repository of circadian omic time series data analyzed using deep learning. Clinicians and researchers can use CircadiOmics to not only validate the findings of their circadian omic experiments, but also to analyze multiple circadian omic experiments in aggregate.

Trusted hardware components are essential when protecting the security of our devices and privacy of our online activities. Several kinds of trusted hardware components are widely available, most notably Trusted Execution Environments (TEEs) and Secure Hardware Tokens. Increasing availability of such hardware prompts a natural question: How can systems benefit from these trusted hardware components? In this dissertation, we design four systems (COMIT, PDoT, CACTI, and VICEROY) that have enhanced security and privacy properties due to the integration of trusted hardware components. We identify and address the key challenges and issues that arise during the integration process. By evaluating proof-of-concept implementations of the four systems, we show that they meet necessary security, privacy, latency, throughput, and deployment requirements.

The United States is currently experiencing its worst drug crisis, which is largely driven by opioid addiction and primarily due to fentanyl. It is therefore necessary to investigate the mechanisms mediating fentanyl's rewarding and reinforcing properties to contribute to the development of successful treatment strategies. Gut bacteria communicate with the brain, and vice versa, via the gut-brain axis to regulate brain function, mood, and behavior. Addiction is a chronic brain disorder that alters circuitry involved in reward, stress, learning, and motivation, all of which have a bidirectional influence between their associated behaviors and gut bacteria. Given the associations between opioid use, gastrointestinal distress, and microbial dysbiosis in humans and rodents, I tested the hypothesis that interactions between gut bacteria and the brain mediate the reinforcing and motivational properties of fentanyl. In this dissertation, I present my work that supports a bidirectional relationship between gut bacteria and fentanyl intravenous self-administration (IVSA) in Sprague Dawley rats. In the following studies, I implanted rats with intravenous catheters in preparation for fentanyl IVSA on an escalating schedule of reinforcement and analyzed gut microbiota by sequencing bacterial DNA from rat fecal samples. I demonstrate that based on sex and fentanyl dose, the diversity of gut bacteria is either increased or decreased following fentanyl IVSA and predicts progressive ratio breakpoint, a measure of motivation. Further, I show that depletion of gut bacteria via prolonged oral antibiotic treatment enhances fentanyl IVSA, and restoration of microbial metabolites with short-chain fatty acid administration decreases fentanyl IVSA back to controls at higher fixed ratio schedules of reinforcement. My findings highlight an important relationship between the knockdown and rescue of gut bacterial metabolites and fentanyl self-administration in adult rats, which provides support for a relationship between gut microbiota and opioid use. Further work in this area may lead to effective, targeted treatment interventions in opioid-related disorders.

My graduate studies in Professor Vy M. Dong’s lab at the University of California, Irvine, have focused on the development of new reaction methodology, with a strong focus toward C–N bond formation. This includes the development of an atom economic hydroamination using feedstock dienes and pyrazoles, the efficient macrocyclization of peptidyl natural products using dehydroamino acids as traceless turn inducers, the design and synthesis of novel photocleavable cross linkers, and the introduction of unique azatricyclic scaffolds as potential benzene mimetics. Through these efforts, we aim to further contribute to the field of organic synthesis, developing robust, cost effective transformations that access synthetically complex building blocks in high efficiency and selectivity.

Contemporary art in all its forms can be a means for the exploration and expression of Blackness and the memorialization of Black humanity. This dissertation project addresses art related to the ocean and connected to expanded notions of Blackness and Black humanity. If the ocean can be said to have memory, then surely it is a forceful one, as vast as its spread over the majority of earth’s surface. During the period when the transatlantic trade in human beings from Africa was extant, millions of Africans did not survive the journey through the Middle Passage and thus remain in the ocean in “residence time” as the term is known in the scientific realm, and as illuminated by the scholar Christina Sharpe in her seminal work, “In The Wake: On Blackness and Being.” The ocean’s memory may contain some evidence of the life force of captive Africans, invisible to us in our contemporary scientific methods of discovery, but present nonetheless. Upon their entry into the sea depths they became part of its universe of sea creatures, microbes and its myriad life-generating processes that populate the ocean, some known to us but many as yet undiscovered. For people of African descent they are also our ancestors, and although they cannot ever be recovered in their complete physicality, there is great value in remembering them and accounting for their existence beyond the bare numerical statistics of the trade. They remain part of Blackness and Black identity, and of our unique history as human beings who facilitated the growth of western democracies with the sweat and suffering of their unpaid labor. We can memorialize them as a strategy against the contested spaces and fraught circumstances in which Black people exist, such that a reaffirmation of Black humanity remains ever necessary.