Open Research Oklahoma (Oklahoma State Univ.)
Not a member yet
42035 research outputs found
Sort by
Exploring deep learning methods for missing data imputation in longitudinal electronic health record data
No full textThis study investigates the efficacy of deep learning methods for the imputation of missing observations in longitudinal electronic health record (EHR) data. Specifically, this study aims to determine if deep learning outperforms traditional multiple imputation methods by handling large, complex temporal data structures without relying on missing data assumptions. We consider deep learning methods commonly used for temporal data structures, including long short-term memory (LSTM) networks and convolutional neural networks (CNNs), for the imputation of missing laboratory and vital sign measurements within a chronic pain EHR study. With comparison to the commonly used multiple imputation by chained equations (MICE) method, we evaluate the imputation performance using root mean square error (RMSE), while monitoring statistical properties including bias, efficiency, and consistency. Additionally, we assess the impact of deep learning-based imputation on subsequent longitudinal analyses. Our results indicate that deep learning methods generally outperform MICE in terms of imputation RMSE. While all imputation methods did not introduce significant bias into imputed results, they all tend to underestimate the variability of the complete data, indicating low imputation precision. Above all, our investigation yields higher prediction accuracy in subsequent longitudinal analyses after applying deep learning-based imputation to the incomplete chronic pain EHR data set. These findings highlight the potential of deep learning in improving imputation accuracy without relying on missing data assumptions required by traditional imputation methods, such as those required by MICE. Ultimately, our study demonstrates the positive impact of deep learning-based imputation on subsequent analyses in longitudinal EHR research, offering an alternative approach to improving statistical analyses in the presence of missing data, and providing more accurate guidance for researchers and healthcare professionals
Planting and care of lawns
No full textThe Oklahoma Cooperative Extension Service periodically issues revisions to its publications. The most current edition is made available. For access to an earlier edition, if available for this title, please contact the Oklahoma State University Library Archives by email at [email protected] or by phone at 405-744-6311
Should I buy (or retain) stockers to graze wheat pasture
No full textThe Oklahoma Cooperative Extension Service periodically issues revisions to its publications. The most current edition is made available. For access to an earlier edition, if available for this title, please contact the Oklahoma State University Library Archives by email at [email protected] or by phone at 405-744-6311
Survey of the perceptions, knowledge, and utilization of core outcome sets in Parkinson’s (PD) trials: A cross-sectional study
No full textBackground: Parkinson’s Disease (PD) is a neurodegenerative condition affecting more than 10 million individuals globally, with an increasing annual incidence and profound, multifactorial consequences. Quality medical decision-making for PD patients is guided by clinical trials. Given the importance of clinical trials, Core Outcome Sets (COS) were created to provide standardized recommendations for trial measurements so trial efficacy could be more accurately compared. However, the PD COS have been poorly implemented in clinical trials since its publication in 2018. As funding for PD research increases, it becomes increasingly critical to comprehend the factors impacting the adoption of PD COS, ensuring the supported research holds clinical significance. This cross-sectional study aims to identify these factors by gathering trialists’ insights in a web-based survey. Our primary objective is to gain a more robust understanding of trialists’ perception, awareness, and experience with the current PD COS to identify its implementation barriers.Methods: In a previous study, we extracted clinical trial measurement tools from PD trials before and after COS publication to evaluate PD COS uptake. For this study, we screened this set of trials to include clinical trialists who have participated in the design, implementation, or analysis of PD trials within the past five years. We then extracted the contact information of 1000 trialists to serve as survey recipients. The survey is designed to be comprehensive and will consist of a set of 20 questions. Participants will have informed consent and maintain complete anonymity. The participants' familiarity with the COS will determine their navigation through the survey. Surveys will be developed and distributed to trialists via REDCap (Research Electronic Data Capture), a secure web-based application designed for research data collection. Data analysis may include both descriptive and inferential statistics. Qualitative data will also be obtained from open-ended questions.Results: Data is currently in the collection phase of this study. Analysis of survey responses will include: (1) Descriptive statistics: summarizes patient demographics and responses to close-ended questions. (2) Inferential statistics: examples include chi-square tests and t-tests which may be used to identify relationships between variables or differences among subgroups. (3) Qualitative data: derived from responses to open-ended questions, which will undergo thematic analysis to discern recurring themes and patterns.Conclusion: Upon completion of this project, our data will inform us of the use and knowledge of COS by clinical trialists. The insight gained from this study may serve as a foundation for future initiatives and interventions aimed at enhancing the utilization of COS among clinical trialists. These outcomes may promote uniformity in clinical COS reporting, ultimately improving patient outcomes with PD
Understanding graphic designers process of making design decisions
No full textUnderstanding graphic designers’ mindsets in how they arrive at solutions to problems they are given is important when analyzing and gaining knowledge in how the behind scenes thought process and driving force impact end results. Even more so, having more impactful knowledge of graphic designers within the agricultural industry.The purpose of this study is to understand better the graphic designer’s side of design within the agricultural industry to be able to follow the pattern of direction in how design results are achieved.Analyzing graphic designers within the agricultural and livestock industry in how they make design decisions. Additionally, it was challenging to find previously conducted research on agricultural graphic designers with making decisions. There is readily available literature on graphic design decision making and the agricultural graphic design. However, a gap of knowledge of and a lack of literature is present on agricultural graphic design decision making and understanding the process. This research seeks to address the gap of knowledge on agricultural graphic designers making creative decisions to better understand the reasoning process
Oklahoma’s promise: A quantitative assessment of engineering, aviation, and nursing graduates in Oklahoma
No full textEngineering, aviation/aerospace, and nursing are top critical occupations in Oklahoma, meaning each area has consistent workforce shortages when compared to the notable industry growth projected over the next ten years (Oklahoma Aeronautics Commission, 2023; Oklahoma Career Tech, 2023; OSRHE, 2021; U.S. Bureau of Labor Statistics, 2023). In higher education and industry research, retention and transition theories are foundational in developing programs and initiatives to motivate selection, pursuit, and career in certain areas of interest. This study discussed the demographics of engineering, aviation/aerospace, and nursing graduates and then evaluated these three degree programs in relation to Oklahoma’s Promise and employment status in Oklahoma. Through descriptive statistics, engineering, aviation/aerospace, and nursing degree program graduates from the 2020-2021 academic year were divided into the following demographics: CIP Codes, residential status, Oklahoma’s Promise recipient status, Pell Grant recipient status, race, gender, and employment status. After discussing demographics, six Chi-Square Tests of Independence were run to determine significant relationships between employment status, degree program, and Oklahoma’s Promise. Results indicated no significant relationship between degree programs and employment status. However, significant relationships were found between degree programs and Oklahoma’s Promise recipient status for those employed in Oklahoma, as well as employment status and Oklahoma’s Promise recipient status for all degree programs. It was concluded that engineering, aviation/aerospace, and nursing graduates who received Oklahoma’s Promise will likely be employed in Oklahoma upon graduation, along with all other degree program graduates who received Oklahoma’s Promise. While Oklahoma’s Promise has proven successful in students’ persistence to graduation and employment, it is not degree-specific and does not appear to significantly impact certain degree programs more than others. Ultimately, large-scale retention incentive programs appear successful, but utilizing more degree-specific programs to meet state-level workforce needs could be beneficial
Holding social media companies accountable for enabling hate and disinformation
No full textSocial media platforms, while connecting billions and amplifying marginalized voices, have become tools for spreading hate, disinformation, and extremist ideologies due to these business models prioritizing engagement and ad revenue.
Engagement-driven algorithms incentivize the spread of harmful content, since inflammatory and divisive posts often garner the most attention, creating a cycle that prioritizes profits over societal well-being.
Social media companies have often been seen as hesitant to enforce their policies against misleading political ads due to the substantial revenue these ads generate. The challenge is further compounded by the high cost of effective content moderation in non-English languages, which creates additional barriers to maintaining platform integrity.
Voluntary self-regulation by social media companies has been inadequate. Governments and international organizations need to step in to enforce meaningful standards for content moderation. Potential approaches include substantial fines for repeated failures, mandatory investment in content moderation, regular third-party audits, and re-examination of legal frameworks to hold companies accountable for algorithmic amplification of harmful content.
The power of social media companies, if unchecked, poses a danger to democratic institutions. The failure to moderate online content can fuel real-world violence, deepen societal divisions, and erode public trust in democracy. Coordinated regional and global efforts are crucial to ensure consistent and effective standards for social media governance.Media and Strategic Communication
Mind is everything, what you think you become: Can mind-body well-being predict job performance?
No full textThe widespread effects of the COVID-19 pandemic on individual and societal physical health and mental well-being have permeated all aspects of life, including business and the workforce. The interaction between stress, illness, well-being, wellness, and performance is complex, including performance in entrepreneurship. The mind-body connection and its impact on wellness and illness has been studied since ancient times and infiltrates centuries-old mindfulness approaches and practices. Modernly, this connection has been researched as the mental state and physiology interaction. This study utilizes (1) machine learning techniques to test models predicting job performance by mind-body measures and (2) multiple regression modeling to investigate the interactions of mind and body measures with each other and with job performance. As the only study examining the relation between mind-body both separately and independently with job performance, the contributions are twofold. First, it advances the literature on mental and physical well-being effects on job performance. Secondly, it not only emphasizes the importance of mental challenges for the employee such as stress and anxiety, but it also highlights specific areas of mental challenges that could be addressed to improve job performance
Foamability properties of low GWP refrigerant and oil mixtures
No full textOil-refrigerant mixture foaming is common in compressors found in many HVAC systems. Low-GWP refrigerants, particularly HFOs and blends, have under-studied foaming behavior and characterizing and understanding foaming behavior is necessary for the integration of new refrigerants in vapor compression cycles. In this work, an apparatus is designed and fabricated to study oil-refrigerant pairs in three manners: 1) physical foaming characterization, 2) measurement of dynamic surface tension between oil-refrigerant media, and 3) measurement of oil-refrigerant solubility data. Foaming is generated through pressure drop by charging and heating a chamber of refrigerant and oil and rapidly connecting it to a low-pressure chamber. Dynamic surface tension data is measured in post-processing via maximum bubble pressure tensiometry (MBPT). Foaming and dynamic surface tension experiments are visually recorded with a custom made, high-pressure sight glass which utilizes pressurized water to attain higher internal oil-refrigerant pressures. Physical characterization of foaming and bubbles is carried out in post-processing with camera-tracking software. Solubility is experimentally determined with a circulation loop containing a viscometer and a densitometer and thermodynamic correlations. The completed apparatus will be used to characterize the behavior of 11 pairs of refrigerant-lubricants including various viscosity grades of mineral oil, POE, PAG, and PVE lubricants paired with HFC and HFO refrigerants
Early termination of cover crops, a multi-degree spline plateau function, and spatial-temporal modeling of crop yield distributions
No full textThe first essay examined the impact of cover crops and termination timing on wheat forage yield, quality, and profitability in Oklahoma. While cover crops did not significantly affect forage yield or nutritional quality, early termination increased yield by 387.7 kg ha⁻¹ over late termination. Economic analysis found the no-till summer fallow system most profitable, with sensitivity analysis indicating cowpea seed costs would need to drop 40% for this system to break even with no-till fallow.
The second essay introduced a multi-degree spline plateau model for optimizing nitrogen fertilizer rates, which is linear in its parameters, simplifying estimation compared to traditional methods. Using data from PRNT, MRTN, and Experiment 502, results indicate that the spline model—with appropriate knot selection—outperforms linear and quadratic plateau models, showing promise for precision agriculture applications. Future research will integrate spatial random effects to enhance predictive accuracy and efficiency.
The third essay estimated crop yield distribution using Bayesian spatial-temporal models for Oklahoma hay and Iowa corn, focusing on the influence of historical data spans on model performance. Both data sets had strong spatial patterns with yields increasing from south-to-north for Iowa corn and increasing from west-to-east for Oklahoma hay. The out-of-sample fit for the spatial models did not exceed the models that ignored spatial information. For the spatial-temporal models, time spans at or near the longest available provided the best fit. When trend was modeled separately, time spans of fifteen years provided the best fit. The results show that spatial information did not serve as a substitute for long time-series of data in modeling crop yield distributions