136879 research outputs found
Sort by
Reducing the Risk of Water Contamination by Improving Livestock Manure Storage and Treatment Facilities
No full textApplication of Causal Inference to Analyze Microseismic and Seismic Events in Unconventional Plays
No full textThis research applies advanced causal inference techniques to uncover intricate causal relationships in two domains within the oil and gas industry - microseismic events and induced seismicity. Microseismic events are small-magnitude seismic events that occur due to crack propagation and rock failure caused by hydraulic fracturing operations. In contrast, induced seismicity refers to larger magnitude earthquakes that occur later in time and over a more extensive area due to subsurface fluid injection operations such as wastewater disposal or hydraulic fracturing. Regarding microseismic events, the study utilizes data from two horizontal wells, MIP 5H and MIP 3H, situated within the Marcellus Shale Energy and Environment Laboratory (MSEEL). Through meticulous spatiotemporal sampling and the formulation of treatment, outcome and confounder variables, Double Machine Learning (DML) is implemented to estimate causal effects. The findings reveal pivotal causal mechanisms governing the characteristics of new microseismic events based on attributes of prior proximal events. High-magnitude prior events are linked to subsequent high-magnitude and delayed new events. Spatial and temporal concentration of prior events also causally impact new event proximity and timing. Appropriate confounder selection is critical, as errors may lead to over/underestimations of the Average Treatment Effect (ATE) by up to 156%.
For induced seismicity, the research explores the causal relationship between injection activities and earthquakes in Oklahoma state using well and seismic data over a 7-year period. Spatiotemporal sampling facilitates the analysis of various grid-time combinations. DML results indicate that approximately 100 water disposal wells induce about 53 earthquakes over 4400 km2 within 54 to 58 days, while 100 hydraulic fracturing wells lead to 36 earthquakes within 16 to 324 km2 over 102 to 110 days. This reveals a more rapid and expansive impact of water disposal versus hydraulic fracturing on induced seismicity. However, minimal causal association is discerned between injection activities and earthquake magnitude.
Overall, the integration of causal inference techniques enhances geoscience data analysis, unearthing genuine causal mechanisms beyond correlational approaches. It strengthens decision-making capabilities regarding extraction practices and mitigation strategies for induced seismicity. This research marks a vital advancement in leveraging causal thinking to unravel the intricate complexities that characterize geophysical systems and phenomena
Exploring the Effect of Changes in Plasma and Organelle Membranes on Alphasynuclein Aggregation
No full textParkinson���s disease (PD) is the second-most common neurodegenerative disorder among seniors after Alzheimer���s disease, affecting 2-3% of the population over 65 years old. While PD typically affects individuals over 55 years old, it can also occur in younger people, known as early-onset PD. Clinical studies have identified Lewy bodies (LBs) in patients��� brain tissue as a hallmark of PD. LBs are intracellular deposits that are primarily composed of a protein called alpha-synuclein (��-Syn). In PD, LBs are present in certain areas of the brain responsible for movement control, such as the substantia nigra pars compacta.
��-Syn is a presynaptic protein that regulates the release of the neurotransmitter dopamine and maintains an adequate supply of synaptic vesicles in presynaptic terminals. ��-Syn is a 140 amino acids protein consisting of 3 regions: N-terminus, non-Amyloid Beta component(NAC), and C-terminus. Under pathological conditions, ��-Syn can aggregate and form toxic oligomers and fibrils, which impair synaptic transmission and cause cell toxicity. The accumulation of these toxic species leads to the formation of LBs.
PD, like other neurodegenerative diseases, is thought to be associated with aging. Aging is a natural and inevitable biological process, accompanied by a deterioration in the body���s function from cellular to organismal. One of the characteristics of normal aging is membrane disruption. The plasma membrane can have alternations in membrane lipid composition, saturation, and membrane fluidity. Some studies have indicated that those alternations could be associated with the onset of ��-Syn aggregation. Therefore, understanding the interaction between ��-Syn and lipid membranes is one of the keys to answering the aggregation mechanisms of ��-Syn and the distribution of ��-Syn aggregates inside the brain. This will allow for connecting the cellular toxicity of ��-Syn aggregates to their structures.
Current methods for analyzing the roles of the lipid membrane and ��-Syn interaction aim to understand the formation of ��-sheet/ ��-strand conformation under the presence of liposome. The formation and accumulation of ��-sheet conformation are known to be associated with the formation of toxic protein aggregates, for example, late-stage oligomers and fibrils. To monitor the ��-sheet content during aggregation, both biochemical, microscopy methods, and other structural analysis methods could be used. Biochemical methods, for example, Thioflavin-T (ThT) fluorescence assay use a fluorescent dye that will interact with the ��-sheet conformation to give a real-time aggregation kinetic. The effect of the presence of liposomes can be reflected by the different takeoff points in the kinetics curves. Microscopy methods such as atomic force microscopy (AFM) and transmissive electron microscopy (TEM) enable the direct observation of different morphologies of protein aggregates. By using AFM or TEM, we can have a visual estimation of the oligomers and fibrils distribution during the aggregation. Spectroscopic methods such as Fournier Transition infrared spectroscopy and Raman spectroscopy can provide protein secondary structure information that is closely related to the structural development during aggregation. Other methods like Solid-state Nuclear Magnetic Resonance (NMR) can reveal the sequence of ��-Syn binding to the lipid membrane, and Cryo-EM could reveal the structure of fibrils and the distance between ��-strands.
��-Syn aggregates, including oligomer and fibril, are known to be toxic to neuron cells. Both ��-Syn oligomers and fibrils can bind to the cell plasma membrane. After binding to the plasma membrane, ��-Syn fibril will form oligomers. These oligomers can also get into neurons through endocytosis or directly permeabilize the lipid membrane. In the cytosol, ��-Syn aggregates induce oxidative stress and cause mitochondrial dysfunction. However, there are many factors to initiate the ��-Syn aggregation. For example, the mutation happens during the transcription or translation, the induced oxidative stress in neurons, or post-translational modification of ��-Syn could all affect the stability of ��-Syn and stimulate the aggregation process. As an age-related neurodegenerative disease, the onset of PD may be related to some age features. One of the changes during the aging process is the change of lipid contents caused by age-related lipid peroxidation in the plasma membrane. We hypothesized that the age-related changes on the plasma membrane could trigger the ��-Syn aggregation process. To elucidate the roles of lipid changes in the plasma membrane during ��-Syn aggregation, we investigated the effects of charge, saturation level, and membrane fluidity properties in ��-Syn aggregation and ��-Syn aggregates cellular toxicity. This information could provide in-depth information for targeting certain lipid pathways for Parkinson���s disease treatment
Prompting LLMs for Zero-Shot Next-Item Recommendation
No full textRecommendation systems can offer personalized suggestions based on user data, enhancing user experiences. Sequential recommendation is a specific subcategory of recommendation systems that focuses on the order and context of previous interactions to offer a sequence of item suggestions. Large language models (LLMs) have demonstrated exceptional prowess in tasks involving commonsense reasoning, knowledge utilization, and task generalization. In particular, they have exhibited remarkable zero-shot performance in numerous natural language processing (NLP) tasks, showcasing their capacity to recommend without extensive training data. This thesis explores the untapped potential of integrating LLMs into sequential recommendations to enhance performance and elevate the user experience. However, this endeavor faces three primary challenges: (i) The huge recommendation space poses significant challenges to LLM-based recommendations, (ii) LLMs face a fundamental constraint that it is not feasible to include all possible items in the prompt, and (iii) for items that appear infrequently in the training of an LLM, it can be challenging to model these items well to make recommendations. In this thesis, we propose a prompting strategy called zero-shot next-item recommendation (NIR) prompting to guide LLM to make next-item recommendations. Specifically, NIR-based strategies involve using external modules to generate candidate projects based on user filtering. Our strategy uses a 4-step prompt to guide GPT-3 to learn the user���s preferences through the user���s past interaction history and recommend a ranked list of K movies. We evaluate the proposed method using GPT-3 on the MovieLens 100K dataset and show that it achieves strong zero-shot performance. Additionally, we also demonstrate that LLMs can be affected by biases like position bias and popularity bias. By employing specialized prompting and bootstrapping strategies, these biases can be mitigated
