Schedule of Events

Previous research has reported that male long-distance runners have normal bone mineral density (BMD) at their hip but low lumbar spine BMD (Fredericson et al., 2007; Hind et al., 2006). An understanding of how the early loading characteristics of ground reaction forces may influence lumbar spine BMD values is lacking. The purpose of this study was to examine the effects of early impact loading rates on hip and lumbar spine BMD in male long-distance runners. Forty injury-free male runners aged 18-30 years ran across a Kistler force plate (1000 Hz) at a self-selected pace with each foot making contact for three trials per foot. Whole-body, hip, and lumbar spine scans were completed with a Hologic dual-energy X-ray absorptiometry machine. Participants who consistently exhibited a heel-striking pattern were divided into two groups (n=16 each) based on their average early loading rate (ELR) values obtained with Bioware software. The Low and High group ELR values were statistically different (86.0 ± 18.3 kN/s vs. 175.3 ± 48.9 kN/s; p < 0.001). The High ELR group had greater BMD values at all anatomical locations analyzed, significantly so at the Ward’s triangle and total lumbar regions (p=0.023 and p=0.002, respectively). The lumbar spine T-scores were significantly different (p=0.003) with 14 subjects in the Low group and 2 subjects in the High group having T scores < -1.0. The results suggest that a higher early impact loading rate may be desirable to obtain adequate BMD for the spine and thus prevent future osteoporic injuries.

Assuming that politicians do in fact lie in some manner, this paper explores the manner in which they lie in order to stay in office. Using definitions and typology from the Oxford Handbook of lying along with data from PolitiFact, we create a model that finds the optimal amount and severity of lies a politican must tell in order to gain the most amount of votes. We also discuss implications of lies, the politicians' considerations when lying, and future models used to better explore this area.

Microfibers are a subset of microplastic that consists of small fibers derived from synthetic clothing and textiles. Microfibers are abundant in the water column, ocean sediments and individuals across all marine phyla, especially filter feeders, are ingesting mircofibers through their feeding strategies. This study aims to determine if there is a difference microfiber uptake within the filter feeding mussel (Mytilus) collected from distinct marine habitats; coastal harbors and exposed jetties. We also want to compare the amount of microfibers within mussels of different sizes. To prepare the mussels for analysis, their shells were measured, had the visceral mass dissected, mashed and submerged into a seawater. This mixture was filtered via suction filtration and examined under a dissection microscope. We found a significant difference in the mean number of microfibers between harbors and jetties (t=-8.66, df= 142.91, p<0.001). We also found that large mussels have more microfibers than smaller mussels.
 

Artificial Neural Networks (ANN) is a mathematical computing system that imitates neurons in the human nervous system, in which ANN collects and classifies information according to a specific structure. They can be trained to recognized patterns in a complex data set, and in particular, their self-learning capabilities enable them to be optimized for better prediction with sufficient accuracy as more relevant data become available. In this project, the ANN platform was created in a MATLAB format for designing the network, assessing network functionality, and optimizing the model factors. The dataset was divided into three models with various divisions of data. Each network model was trained and compared to experimental results under a supervised learning model using a two-layer feed-forward network with one hidden layer, from three up to fifteen hidden neurons. Subsequently, the performance of all models was observed under the correlation coefficient as well as the model error and discrepancies between predictive and experimental results. The model had ten hidden neurons with 70% data for training, 30% data for validating, and testing had the most remarkable performance as it had a high correlation as well as a small discrepancy in predictive result with a total of seven iterations for validation and four iterations for error validation. Therefore, an optimized ANN model will prove its accuracy and robustness when deploying them to assess the toxicity of phenolic-related compounds via chemical structure and function. Furthermore, they will predict toxicity values for human health assessment and industrial worker exposure.
 

After the Higgs Boson was detected in the Large Hadron Collider (LHC), models that could describe the process of creation of Higgs particles where theorized and tested. One mechanism includes Vector Boson Fusion (VBF) production where the Higgs is created by the fusion of two W or Z boson virtual particles. We are studying the decay of a pair of VBF Higgs into four muons. This research project will develop an algorithm that can separate Monte Carlo samples of these VBF Higgs events from its background processes such as quark diffractive and ZZ boson Drell-Yan production. The aim is to enrich our sample data selection schemes while reducing the statistical contamination from these two background sources. This work was done in collaboration with student Connor Hinthorn. In particular, my part of this collaborative project is to develop a machine learning algorithm that can accomplish the increased statistical significance of the signal process. The data analysis on the samples is done with software we have collaboratively developed that uses the CERN CMS experiment standard Python and C++ based software package. Much of the work to condition our data analysis software was done, as well as setting up our algorithm framework. We have produced and filtered both signal VBF Higgs and background ZZ samples. We have also settled on the data analysis event selection criteria. We will present this work as well as our initial efforts to implement a Neural Network signal/background separation algorithm on our simulated signal and background filtered samples.