This study explores the application of fine-tuned large language models for predicting physicochemical properties, specifically focusing on Abraham model solute descriptors (E, S, A, B, V) and modified solvent parameters (e0, s0, a0, b0, v0). By leveraging ChemLLaMA, a specialized version of the LLaMA model for cheminformatics tasks, we developed the AbraLlama-Solvent and AbraLlama-Solute models using curated datasets of experimentally derived solute descriptors and solvent parameters. Our findings demonstrate that AbraLlama-Solvent and AbraLlama-Solute predict modified solvent parameters and solute descriptors with high accuracy, comparable to existing methods. The AbraLlama-Solvent model shows varying prediction accuracy across different solvents, influenced by their position within the chemical space, while the AbraLlama-Solute model consistently predicts solute descriptors with high accuracy. Both models are available as applications on Hugging Face, facilitating easy predictions from SMILES strings. This research highlights the potential of LLMs in chemistry applications, offering practical tools for solvent comparison and expanding the applicability of Abraham solvation equations to a broader range of organic solvents.

This study introduces a systematic framework to compare the efficacy of Large Language Models (LLMs) for fine-tuning across various cheminformatics tasks. Employing a uniform training methodology, we assessed three well-known models-RoBERTa, BART, and LLaMA-on their ability to predict molecular properties using the Simplified Molecular Input Line Entry System (SMILES) as a universal molecular representation format. Our comparative analysis involved pre-training 18 configurations of these models, with varying parameter sizes and dataset scales, followed by fine-tuning them on six benchmarking tasks from DeepChem. We maintained consistent training environments across models to ensure reliable comparisons. This approach allowed us to assess the influence of model type, size, and training dataset size on model performance. Specifically, we found that LLaMA-based models generally offered the lowest validation loss, suggesting their superior adaptability across tasks and scales. However, we observed that absolute validation loss is not a definitive indicator of model performance - contradicts previous research - at least for fine-tuning tasks: instead, model size plays a crucial role. Through rigorous replication and validation, involving multiple training and fine-tuning cycles, our study not only delineates the strengths and limitations of each model type but also provides a robust methodology for selecting the most suitable LLM for specific cheminformatics applications. This research underscores the importance of considering model architecture and dataset characteristics in deploying AI for molecular property prediction, paving the way for more informed and effective utilization of AI in drug discovery and related fields.

This paper examines the transformative impact of Artificial Intelligence (AI) tools, especially Large Language Models like ChatGPT, on higher education. Focusing on how AI can enhance and challenge the learning environment, it navigates through the benefits and ethical concerns, such as privacy issues, overreliance on the technology itself, and potential biases. The article's core comprises two practical case studies-one in computer science, where ChatGPT aids in teaching programming, and another in English composition, exploring its role in developing writing skills. In the computer science context, ChatGPT shows how AI can introduce diverse problem-solving approaches and elevate student engagement, with notable improvements in students' comprehension and application of programming techniques. In English composition, the integration of ChatGPT assists in crafting texts, highlighting the balance needed between AI assistance and human critical thinking. Concluding with a call for a balanced approach, the study emphasizes that AI should complement, not substitute, traditional teaching methods. It advocates for a responsible and ethical application of AI in education, underlining the need to integrate technological advancements with fundamental core literacies to elevate the academic experience of all students.

Background : This study examined the relationship between physical activity (PA) and academic performance and retention among college students using accelerometer data while controlling for sex and socioeconomic background. Methods : Data were collected from 4643 fi rst-year college students at a private university in the south-central United States who began their studies in the Fall semesters between 2015 and 2022. Daily step counts were collected using accelerometers worn as part of the institutions PA requirements. The expected family contribution was calculated based on information provided on the Free Application for Federal Student Aid. Grade point average (GPA) and retention data were collected directly from of fi cial university databases. Results : Female students generally had lower median age and steps count and a higher median GPA than males. The retention rates from fall to spring and fall to fall were 95.9% and 83.8%, respectively, with no signi fi cant difference in retention rates between males and females. GPA was signi fi cantly positively correlated with steps, expected family contribution, and age, and negatively correlated with being male and having an expected family contribution of zero. The fall to spring retention rate was signi fi cantly positively correlated with steps and GPA. Conclusions : The study ' s fi ndings suggest that increasing levels of PA, as measured by daily step counts, may be effective in promoting academic performance and retention, even when controlling for sex and socioeconomic background. The use of device-based measures of PA in this study contributes to the literature on this topic, and policymakers and educational institutions should consider interventions focused on academic performance and physical activity to help students persist.

The real-time quantification of the effect of a wireless channel on the transmitting signal is crucial for the analysis and the intelligent design of wireless communication systems for various services. Recent mechanisms to model channel characteristics independent of coding, modulation, signal processing, etc., using deep learning neural networks are promising solutions. However, the current approaches are neither statistically accurate nor able to adapt to the changing environment. In this paper, we propose a new approach that combines a deep learning neural network with a mixture density network model to derive the conditional probability density function (PDF) of receiving power given a communication distance in general wireless communication systems. Furthermore, a deep transfer learning scheme is designed and implemented to allow the channel model to dynamically adapt to changes in communication environments. Extensive experiments on Nakagami fading channel model and Log-normal shadowing channel model with path loss and noise show that the new approach is more statistically accurate, faster, and more robust than the previous deep learning-based channel models.

In the realm of wireless communication, stochastic modeling of channels is instrumental for the assessment and design of operational systems. Deep learning neural networks (DLNN), including generative adversarial networks (GANs), are being used to approximate wireless Orthogonal frequency-division multiplexing (OFDM) channels with fading and noise, using real measurement data. These models primarily focus on channel output (y) distribution given input x: p(y|x), limiting their application scope. DLNN channel models have been tested predominantly on simple simulated channels. In this paper, we build both GANs and feedforward neural networks (FNN) to approximate a more general channel model, which is represented by a conditional probability density function (PDF) of receiving signal or power of node receiving power Prx: f_p_rx|d(()), where is communication distance. The stochastic models are trained and tested for the impact of fading channels on transmissions of OFDM QAM modulated signal and transmissions of general signal regardless of modulations. New metrics are proposed for evaluation of modeling accuracy and comparisons of the GAN-based model with the FNN-based model. Extensive experiments on Nakagami fading channel show accuracy and the effectiveness of the approaches.

Educational literature has underscored the importance of higher education attending to students’ religious development. For Christian students, environment plays a role. Recent Christian religious development literature focusing on environmental predictors of growth found that higher religious pressures, assumed to be a controller according to Self-Determination Theory, predicted religious growth for students. The current study examines this finding by considering the influence of religious pressures along with normative experiences of religious doubt on spiritual/religious development variables of relatedness and self-mastery. A web-based survey procured responses from a large sample of students from both Christian and public/secular institutions. Controlling for sex and institution, multiple linear regression modeling was used to develop a model hypothesizing that when religious doubts are high, greater pressure would result in lower levels of religious development. The model was supported for spiritual/religious self-mastery but not for relatedness. Students’ quotes were presented to illustrate the findings that emerged from the data analysis. Results clarify the deleterious role of religious pressures for Christians at certain developmental and situational milestones who are also simultaneously experiencing religious doubt and/or questioning their beliefs about God. Given past findings about the uncomfortable and unsatisfying, albeit necessary, role of engaging in some religious doubt and exploration, religious pressures can sabotage effective adaptation. The results underscore the importance of higher education administrators allowing space and support for religious questioning and doubt. Moreover, administrators in Christian universities should help facilitate the honest expression of their students’ doubts and questions and the minimization of environmental religious pressures.

The concept of “solvent polarity” is widely used to explain the effects of using different solvents in various scientific applications. However, a consensus regarding its definition and quantitative measure is still lacking, hindering progress in solvent-based research. This study hopes to add to the conversation by presenting the development of two linear regression models for solvent polarity, based on Reichardt’s ET(30) solvent polarity scale, using Abraham solvent parameters and a transformer-based model for predicting solvent polarity directly from molecular structure. The first linear model incorporates the standard Abraham solvent descriptors s, a, b, and the extended model ionic descriptors j+ and j−, achieving impressive test-set statistics of R2 = 0.940 (coefficient of determination), MAE = 0.037 (mean absolute error), and RMSE = 0.050 (Root-Mean-Square Error). The second model, covering a more extensive chemical space but only using the descriptors s, a, and b, achieves test-set statistics of R2 = 0.842, MAE = 0.085, and RMSE = 0.104. The transformer-based model, applicable to any solvent with an associated SMILES string, achieves test-set statistics of R2 = 0.824, MAE = 0.066, and RMSE = 0.095. Our findings highlight the significance of Abraham solvent parameters, especially the dipolarity/polarizability, hydrogen-bond acidity/basicity, and ionic descriptors, in predicting solvent polarity. These models offer valuable insights for researchers interested in Reichardt’s ET(30) solvent polarity parameter and solvent polarity in general.

Past research has confirmed the utility of environmental variables, and perceptions of religious pressure (RP) in particular, in predicting faith maturity and religious schema scores for participants from Christian environments. Whether environmental variables predict religious development and whether religious development, in turn, leads to greater well-being for individuals from broader environments remain unknown. Utilizing participants from both Christian and non-Christian environments, the current study measures religious development variables that were constructed based on self-determination theory (SDT). We used structural equation modeling (SEM) to evaluate our hypothesis that religious pressures RP and autonomy supportive environment (ASE) are antecedents to religious/spiritual relatedness (R/S-R) and self-mastery (R/S-S), which in turn lead to greater well-being, as determined by the presence of meaning in life (MIL). Results indicate that both environmental variables of RP and ASE predicted higher scores on religious/spiritual relatedness and self-mastery, and this led to self-reports of greater well-being for both samples. Therefore, members of the broader religious environment of Christianity responded to RP similarly, implying that certain commonalities may shape Christians’ cognitions around obedience to God and authority across settings, although this result should be interpreted with caution. Further implications of these findings are explored and recommendations for future research provided.

Annals of Chemical Science Research Fine-Tuning ChemBERTa-2 for Aqueous Solubility Prediction Andrew SID Lang1*, Wei Khiong Chong2 and Jan HR Wörner1 1Department of Computing & Mathematics, Oral Roberts University, USA 1Advent Polytech Co., Ltd., Taipao City, Taiwan *Corresponding author:Andrew SID Lang, Department of Computing & Mathematics, Oral Roberts University, Tulsa, OK, 74171, USA Submission: May 08, 2023;Published: May 19, 2023 DOI: 10.31031/ACSR.2023.04.000578 Volume4 Issue1May , 2023

BackgroundPersonality traits are known factors that may influence levels of physical activity and other healthy lifestyle measures and behaviors that ultimately lead to health problems later in life.Participants and procedureThe aim of this study was to examine the association between personality traits (HEXACO) and levels of physical activity and resting heart rate (RHR) – measured using Fitbits, BMI, and a self-reported whole-person healthy lifestyle score for N = 2580 college students. Data were collected and analyzed for students enrolled in a University Success type course from August 2017 to May 2021. The relationships between HEXACO personality traits and various physical activity and healthy lifestyle behaviors were analyzed by building several multiple regression models using R version 4.0.2.ResultsIn general, students who are extraverted were more physically active and students who are more open to experience had a higher RHR, even when controlling for gender. Females and males however had different profiles as to how personality influenced physical activity and other health-related measures. Male extraverts with high negative emotionality scores tend to be more physically active, whereas females tend to be more physically active when they were high in extroversion and conscientiousness, and low in openness to experience. BMI values were higher for female participants with high honesty-humility and low agreeableness and conscientiousness scores. Females also had a lower RHR for high honesty-humility and emotionality and low conscientiousness scores.ConclusionsPersonality can influence levels of physical activity, RHR, and BMI. This is especially true of women. Being aware of one’s personality and the relationship of personality traits to levels of physical activity and other measures of leading a healthy lifestyle can be beneficial in determining strategies to improve long-term health outcomes.

This paper describes a neural network model that can be used to detect at-risk students failing a particular course using only grade book data from a learning management system. By analyzing data extracted from the learning management system at the end of week 5, the model can predict with an accuracy of 88% whether the student will pass or fail a specific course. Data from the grade books from all course shells from the Spring 2022 semester (N = 22,041 rows) were analyzed, and four factors were found to be significant predictors of student success/failure: the current course grade after the fifth week of the semester and the presence of missing grades in weeks 3, 4, and 5. Several models were investigated before concluding that a neural network model had the best overall utility for the purpose of an early alert system. By categorizing students who are predicted to fail more than one course as being generally at risk, we provide a metric for those who use early warning systems to target resources to the most at-risk students and intervene before students drop out. Seventy-four percent of the students whom our model classified as being generally at risk ended up failing at least one course.

An N = 14,076 dataset consisting of non-athlete students with instructor monitored measured field-test times (1-mile, 1.5-mile, 2-mile).

Purpose. The purpose of this study is to investigate if leading a physically active and healthy lifestyle can prevent the weight gain typically experienced during the freshman year of college – the ‘Freshman 15’. Methods. Study participants (N = 525) were from three cohorts of incoming students (2018–2020) at a mid-sized university in the West South-Central United States. The weight of each study participant was measured at three points over a year: beginning of their first semester, beginning of their second semester, and beginning of their third semester. During the study, students were encouraged to lead physically active and healthy lifestyles and to exercise daily. Weight changes at sixth months and one-year intervals were recorded as percentages. Results. Freshmen weight gain/loss depended upon initial weight with freshmen who arrived on campus with relatively lower weights (≤100 kg) tending to gain weight, especially males, whilst freshmen who arrived on campus with relatively higher weights (≥100 kg) tended to lose weight. This finding was both more apparent and more statistically significant at the 1-year mark than at the 6-months mark. Conclusions. Several previous studies have linked freshmen weight gain to initial weight with students with higher initial weights gaining the most. However, our results show that maintaining a physically active and healthy lifestyle when entering college reverses this trend – with students with high initial weights losing weight. Thus, living a physically active and healthy lifestyle, which includes aerobic exercise, can prevent the fat mass weight gain often experienced by college freshmen.

Recent Christian religious development literature focusing on environmental predictors of growth found that higher religious pressures, assumed to be a controller according to Self-Determination Theory, predicted religious growth. The current study examines this finding by considering the influence of religious pressures along with normative experiences of religious doubt on spiritual/religious development variables of relatedness and self-mastery. A web-based survey procured responses from a large sample of participants from both Christian and secular institutions. Controlling for sex and institution, multiple linear regression modeling was used to develop a model hypothesizing that, when religious doubts are high, greater pressure would result in lower levels of religious development. The model was supported for spiritual/religious self-mastery but not for relatedness. Participants’ quotes were presented to illustrate the findings that emerged from the data analysis. Results clarify the deleterious role of religious pressures for Christians at certain developmental and situational milestones who are also simultaneously experiencing religious doubt and/or questioning their beliefs about God. Given past findings about the uncomfortable and unsatisfying, albeit necessary, role of engaging in some religious doubt and exploration, religious pressures can sabotage effective adaptation. Results underscore the importance of leaders and mentors in Christian environments allowing honest expression of doubts and questions and minimizing environmental religious pressures.

The SARS-CoV-2 Delta variant rose to dominance in mid-2021, likely propelled by an estimated 40%???80% increased transmissibility over Alpha. To investigate if this ostensible difference in transmissibility is uniform across populations, we partner with public health programs from all six states in New England in the United States. We compare logistic growth rates during each variant???s respective emergence period, finding that Delta emerged 1.37???2.63 times faster than Alpha (range across states). We compute variant-specific effective reproductive numbers, estimating that Delta is 63%???167% more transmissible than Alpha (range across states). Finally, we estimate that Delta infections generate on average 6.2 (95% CI 3.1???10.9) times more viral RNA copies per milliliter than Alpha infections during their respective emergence. Overall, our evidence suggests that Delta???s enhanced transmissibility can be attributed to its innate ability to increase infectiousness, but its epidemiological dynamics may vary depending on underlying population attributes and sequencing data availability.

This study quantifies the relationship between measures of spiritual intelligence and personality traits. A random sample of 240 undergraduate students from a mid-sized private Christian university in the West South-Central United States were administered both the SISRI-24 survey instrument for spiritual intelligence as well as the Pathway U survey for HEXACO personality traits. Statistically significant positive correlations were found between critical existential thinking and openness to experience; personal meaning production and extraversion, agreeableness, and conscientiousness; transcendental awareness and extraversion, agreeableness, openness to experience, and conscientiousness; and conscious state expansion and extraversion, agreeableness, openness to experience, and conscientiousness. Statistically significant negative correlations were found between negative emotionality and personal meaning production, transcendental awareness, and conscious state expansion. Interestingly, honesty- humility was found to be significantly negatively correlated with overall SI-score, personal meaning production, transcendental awareness, and conscious state expansion in a multi linear regression analysis but was not a significant factor when taken alone. These results inform those who wish to develop spiritual intelligence by taking individual personality traits into account.

The use of equivalent alkane carbon numbers (EACN) to characterize oils is important in surfactant-oil-water (SOW) systems. However, the measurement of EACN values is non-trivial and thus it becomes desirable to predict EACN values from structure. In this work, we present a simple linear model that can be used to estimate the EACN value of oils with known Abraham solute parameters. We used linear regression with leave-one-out cross validation on a dataset of N = 80 oils with known Abraham solute parameters to derive a general model that can reliably estimate EACN values based upon the Abraham solute parameters: E (the measured liquid or gas molar refraction at 20 °C minus that of a hypothetical alkane of identical volume), S (dipolarity/polarizability), A (hydrogen bond acidity), B (hydrogen bond basicity), and V (McGowan characteristic volume) with good accuracy within the chemical space studied (N = 80, R2 = 0.92, RMSE = 1.16, MAE = 0.90, p < 2.2 × 10−16). These parameters are consistent with those in other models found in the literature and are available for a wide range of compounds.

Background: The coronavirus disease 2019 (COVID-19) pandemic adversely disrupted university student educational experiences worldwide, with consequences that included increased stress levels and unhealthy sedentary behavior. Aim: This study aimed to quantify the degree of impact that COVID-19 had on the levels of physical activity and stress of university students by utilizing wearable fitness tracker data and standard stress survey instrument scores before and during the pandemic. Methods: We collected Fitbit heart rate and physical activity data, and the results of a modified Social Readjustment Rating Scale (SRRS) stress survey from 2,987 university students during the Fall 2019 (residential instruction; before COVID-19) and Fall 2020 (hybrid instruction; during COVID-19) semesters. Results: We found indicators of increased sedentary behavior during the pandemic. There was a significant decrease in both the levels of physical activity as measured by mean daily step count (↓636 steps/day; p = 1.04 · 10-9) and minutes spent in various heart rate zones (↓58 minutes/week; p = 2.20 · 10-16). We also found an increase in stressors during the pandemic, primarily from an increase in the number of students who experienced the “death of a close family member” (38.8%), with the number even higher for the population of students who opted to stay home and attend classes virtually (41.4%). Conclusions: This study quantifies the decrease in levels of physical activity and notes an increase in the number of students who experienced the death of a close family member, a known stressor, during the first year of the COVID-19 pandemic. These findings allow for more informed student-health-focused interventions related to the COVID-19 pandemic disruptions experienced by academic communities worldwide.

The severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) Delta variant quickly rose to dominance in mid-2021, displacing other variants, including Alpha. Studies using data from the United Kingdom and India estimated that Delta was 40-80% more transmissible than Alpha, allowing Delta to become the globally dominant variant. However, it was unclear if the ostensible difference in relative transmissibility was due mostly to innate properties of Delta's infectiousness or differences in the study populations. To investigate, we formed a partnership with SARS-CoV-2 genomic surveillance programs from all six New England US states. By comparing logistic growth rates, we found that Delta emerged 37-163% faster than Alpha in early 2021 (37% Massachusetts, 75% New Hampshire, 95% Maine, 98% Rhode Island, 151% Connecticut, and 163% Vermont). We next computed variant-specific effective reproductive numbers and estimated that Delta was 58-120% more transmissible than Alpha across New England (58% New Hampshire, 68% Massachusetts, 76% Connecticut, 85% Rhode Island, 98% Maine, and 120% Vermont). Finally, using RT-PCR data, we estimated that Delta infections generate on average ∼6 times more viral RNA copies per mL than Alpha infections. Overall, our evidence indicates that Delta's enhanced transmissibility could be attributed to its innate ability to increase infectiousness, but its epidemiological dynamics may vary depending on the underlying immunity and behavior of distinct populations.