Electronic Thesis or Dissertation

Animals display a remarkable variety of social behaviors that are necessary for survival. Despite the importance of social behaviors, the neurobiological mechanisms underlying the evolution of such behaviors are largely unknown. The Mexican tetra, Astyanax mexicanus, is a powerful model for studying how behaviors evolve, including social behavior. A. mexicanus exists as a schooling surface form and a non-schooling cave form. Here we have utilized this model in order to investigate how differences in the behavior of individuals result in differences at the level of emergent group social behaviors. We begin by reviewing how fish have contributed to the study of social behavior in Chapter 1, then continue to dissect differences in the schooling and shoaling behavior of adult surface and cave fish in Chapter 2, and finally address ontogenic differences that result in these differences in Chapter 3. All-in-all this, work reveals how evolution may act on the behavior of individuals to produce differences in relevant group behaviors.

The amyloid beta (Aβ) peptide has been linked to Alzheimer’s Disease (AD) since the early 1990s. Since then, many studies have characterized the peptide and examined its aggregation process. Aβ is a 40 or 42-residue peptide, composed of a charged N-terminal and hydrophobic C-terminal, that aggregates into characteristic β-sheets forming insoluble plaques in the brains of (AD) patients. In recent years an intermediate oligomeric species has been shown to interact with lipid membranes, largely resulting in the etiology of AD. In this study, two fragments are used, the 23-residue N-terminal fragment, Aβ23 and the 30-residue C-terminal fragment, Aβ11-40, to better understand the role of the N and C-terminus in the aggregation of Aβ peptide. Aβ11-40 has also been found in the brains of AD patients, playing a biological role in the disease. This study used analytical and biophysical techniques to systematically synthesize, purify, characterize, and study these fragments' aggregation in different conditions. We investigated the effects of lipid membranes on the aggregation of Aβ23 and Aβ11-40 and the activities of these peptides in inducing membrane damage. The results show that the aggregation of Aβ23 was increased in the presence of lipid membranes, likely due to favorable electrostatic interactions. However, the aggregation of Aβ11-40 was not influenced by lipid membranes. A dye leakage study was carried out to study the membrane damage occurring as a result of fragments' interaction with lipid membranes. The results showed that neither fragment had a profound effect on membrane destruction, although the charge of the lipid head seemed to play a role. This work's second study focused on the effect of three specific polysaccharides, heparin, chitosan (CHT), and trimethyl chitosan (TMC), on the aggregation of Aβ23 and Aβ11-40. The results showed that for Aβ23, heparin increased aggregation, while both CHT and TMC decreased aggregation. However, for Aβ11-40, both heparin and CHT did not affect aggregation, while TMC decreased aggregation.

Health disparities in the US Health care system are a well-known fact. I examined such disparity with an anthropological lens, focusing on how Bangladeshi uninsured and underinsured immigrants navigate the system of doctors, clinics, hospitals, and payment regimes (insurance or not). I focused on how these immigrants experience the American system, how they react to it, interpret it, understand it, and contextualize it from their particular backgrounds and expectations. This study will be a step toward closing the knowledge gap of a particular immigrant group's everyday experience of access to health care in the U.S. This research emphasizes Bangladeshi immigrants' everyday sufferings, their struggle, their anxiety, and frustration with access to U.S. health care services. Besides, this is an opportunity to discover the barriers to healthcare access for Bangladeshi uninsured and underinsured immigrant groups. This study provides as much helpful information as possible about the health-seeking practices of uninsured and underinsured Bangladeshi immigrants through ethnographic experience. This study also shows how poor or low-income people are the victims of a country's structural violence. Furthermore, low-income, uninsured, and underinsured immigrants suffer a lot due to problems in the system. And this study also focuses on holistically understanding social inequalities in healthcare services in the U.S.

Artificial reefs are coastal structures built to improve marine life and prevent beach erosion. During earlier days artificial reefs were constructed for recreational fishing using discarded scraps and waste materials. Later on, ships were scuttled for constructing artificial reefs. Artificial reefs dissipate the energy of the wave by making the wave break over the reef. The artificial reefs used for coastal protection are usually in submerged condition as this condition does not affect the aesthetic beauty of the beach. Wave transmission decides the efficiency of submerged-detached artificial reef in protecting the beach from the incoming waves. The efficiency of submerged detached coastal protection structures in protecting the beach is usually measured in terms of wave transmission coefficient.
The experimental investigation in the present study is carried out for submerged two-dimensional impermeable and permeable reefs for three water depths. The crest width of the reefs considered for the experimental studies are 60 cm and 20 cm. The permeable artificial reefs are made up of oyster shells in Nylon bags and biodegradable bags. The water levels considered for the study are 35 cm, 34 cm, and 33 cm. The effect of pore space between the oyster shells, crest width, water depth and wave parameters on the wave transmission coefficient for submerged impermeable and permeable artificial reefs are studied experimentally. The wave transmission coefficient is calculated for submerged impermeable and permeable reefs for different water levels and crest widths. Based on the results of the present experimental studies, it is logical to conclude that both submerged impermeable and permeable artificial reefs contribute to a significant extent to the attenuation of the incident wave.

The study of the electrical properties of red blood cells (RBCs) plays a crucial role in advancing our understanding of human health. As RBCs age, they undergo changes that affect hemorheology and blood microcirculation, which have far-reaching implications for disease research. Furthermore, the shortage of RBC storage units can be a major issue for patients, underscoring the importance of characterizing RBC aging with respect to cell densities. In individuals with abnormal hemoglobin disease, alterations in hemoglobin and its functionality can modify the volume and density of RBCs, making their study even more crucial. To this end, our aim is to investigate the impedance alterations of RBCs after distributing them into different layers based on their densities. We have developed a novel method for non-invasive, rapid, and real-time single-cell analysis of RBCs. Our approach involves the use of electrical impedance spectroscopy (EIS) to study the cells after performing cell fractionation. Our studies indicate an increasing trend for RBC resistance and a decreasing trend for the cell membrane as the density of the layer increases. Additionally, we have developed a method for extracting hemoglobin with high purity from fresh samples of RBCs. By passing lysed RBCs through ultrafiltration devices and removing debris and membranes, we were able to isolate hemoglobin. Using the EIS technique, we studied the alterations of impedance over a frequency range, obtaining valuable insight into the electrical properties of hemoglobin.

Coastal wetlands across the Indian River Lagoon (IRL) on the east coast of Florida have been impounded for mosquito control purposes, which have been known to have adverse effects on overall fish populations. The objective of this project was to assess the use of culverts by species of larval fish at three impounded mangrove sites in the IRL. Light traps were used to collect samples of larval fish (both inside the basins and in the surrounding lagoon) which were humanely euthanized, preserved, and examined under a digital microscope. A total of 3,926 fish were collected from 24 taxa in 576 samples over the year-long study. Larval seasonality generally followed known reproductive seasonality of the species captured. Inside the impoundments were dominated by species known to spawn in and around mangroves such as the Gambusia holbrooki and Poecilia latipinna. Species that spawn in the IRL or in coastal waters that subsequently use the IRL as a nursery (such as Anchoa mitchilli and Gobiosoma robustum) had relatively low catches in the impoundments. Larvae of the main sportfishery species that have juveniles known to utilize the studied impoundments (Megalops atlanticus, the Atlantic tarpon, and Centropomus undecimalis, the common snook), were rarely caught inside the impoundments or in the surrounding IRL. The low numbers of IRL and offshore spawning larvae that enter the impoundments may be hindered by restricted water flow through culverts connecting the habitats, or by their inability to survive the low DO conditions often found inside the impoundments, especially during the summer. The lack of larval tarpon and snook in the light collections suggest that these species metamorphose from the larval to juvenile stage outside of the impoundments, before they enter the mangrove-dominated nursery habitats. The results of the study can be used to further modify impoundment restoration and management strategies to enhance their role as fish nursery habitats.

The major objective of this dissertation was to create a framework which is used for medical image diagnosis. In this diagnosis, we brought classification and diagnosing of diseases through an Artificial Intelligence based framework, including COVID, Pneumonia, and Melanoma cancer through medical images. The algorithm ran on multiple datasets. A model was developed which detected the medical images through changing hyper-parameters.
The aim of this work was to apply the new transfer learning framework DenseNet-201 for the diagnosis of the diseases and compare the results with the other deep learning models. The novelty in the proposed work was modifying the Dense Net 201 Algorithm, changing hyper parameters (source weights, Batch Size, Epochs, Architecture (number of neurons in hidden layer), learning rate and optimizer) to quantify the results. The novelty also included the training of the model by quantifying weights and in order to get more accuracy. During the data selection process, the data were cleaned, removing all the outliers. Data augmentation was used for the novel architecture to overcome overfitting and hence not producing false absurd results the computational performance was also observed. The proposed model results were also compared with the existing deep learning models and the algorithm was also tested on multiple datasets.

Traditional techniques of observing cracking within reinforced structures can be invasive, leading to an increased risk of added corrosion to structures already undergoing corrosive processes. The research presented in this document improves upon a nondestructive method for detecting early crack formation in reinforced concrete. This method includes using acoustic signaling to add a layer of salt water between the sensor and analyzed sample. Following the collection of surface and rebar echo responses, an adapted version of the novel Biot-Stoll method is used to model sound propagation for poro-elastic mediums. Testing of model parameters and variables has improved the root mean square error (RMSE) by up to 63.7% when studying the full signal, and up to 62.6% for the rebar echo locations. These improvements signify better curve fitting between simulated and measured responses, which lead to increased accuracy in the model parameter outputs.

As the global population is increasing, the generation of various waste materials (fats, oils and grease, fruit waste etc.) is increasing, which when landfilled, takes up valuable landfill space. Anaerobic digestion techniques have been developed that potentially convert these waste materials into energy and fertilizer, thus reducing landfill demand. It has been hypothesized that addition of high strength organic waste to conventional wastewater sludge can enhance the generation of onsite biogas at wastewater treatment plants, to meet the energy requirements of the plant partially or fully.
To determine the anaerobic biodegradability of fats, oils and grease and fruit waste residuals, lab scale ultimate digestibility tests were conducted for a period of 63 days under mesophilic conditions. High strength organic wastes, thickened waste activated sludge and inoculum were mixed at 9 different ratios, and the mixtures were incubated in 500 mL serum bottles. After 63 days, the highest methane yield of 280 mL/gVS and 243 mL/gVS were obtained with mixtures containing 10% FOG with 10% red apples and 10% FOG only respectively whereas the methane yield of inoculum was only 8 mL/gVS. Preliminary cost analyses were conducted using the laboratory derived data

Recent successes of Deep Learning-powered AI are largely due to the trio of: algorithms, GPU computing, and big data. Data could take the shape of hospital records, satellite images, or the text in this paragraph. Deep Learning algorithms typically need massive collections of data before they can make reliable predictions. This limitation inspired investigation into a class of techniques referred to as Data Augmentation. Data Augmentation was originally developed as a set of label-preserving transformations used in order to simulate large datasets from small ones. For example, imagine developing a classifier that categorizes images as either a “cat” or a “dog”. After initial collection and labeling, there may only be 500 of these images, which are not enough data points to train a Deep Learning model. By transforming these images with Data Augmentations such as rotations and brightness modifications, more labeled images are available for model training and classification! In addition to applications for learning from limited labeled data, Data Augmentation can also be used for generalization testing. For example, we can augment the test set to set the visual style of images to “winter” and see how that impacts the performance of a stop sign detector.
The dissertation begins with an overview of Deep Learning methods such as neural network architectures, gradient descent optimization, and generalization testing. Following an initial description of this technology, the dissertation explains overfitting. Overfitting is the crux of Deep Learning methods in which improvements to the training set do not lead to improvements on the testing set. To the rescue are Data Augmentation techniques, of which the Dissertation presents an overview of the augmentations used for both image and text data, as well as the promising potential of generative data augmentation with models such as ChatGPT. The dissertation then describes three major experimental works revolving around CIFAR-10 image classification, language modeling a novel dataset of Keras information, and patient survival classification from COVID-19 Electronic Health Records. The dissertation concludes with a reflection on the evolution of limitations of Deep Learning and directions for future work.