An Explainable Prediction for Dietary-Related Diseases via Language Models

Our study harnesses the power of natural language processing (NLP) to explore the relationship between dietary patterns and metabolic health outcomes among Korean adults using data from the Seventh Korea National Health and Nutrition Examination Survey (KNHANES VII). Using Latent Dirichlet Allocation (LDA) analysis, we identified three distinct dietary patterns: "Traditional and Staple", "Communal and Festive", and "Westernized and Convenience-Oriented". These patterns reflect the diversity of dietary preferences in Korea and reveal the cultural and social dimensions influencing eating habits and their potential implications for public health, particularly concerning obesity and metabolic disorders. Integrating NLP-based indices, including sentiment scores and the identified dietary patterns, into our predictive models significantly enhanced the accuracy of obesity and dyslipidemia predictions. This improvement was consistent across various machine learning techniques-XGBoost, LightGBM, and CatBoost-demonstrating the efficacy of NLP methodologies in refining disease prediction models. Our findings underscore the critical role of dietary patterns as indicators of metabolic diseases. The successful application of NLP techniques offers a novel approach to public health and nutritional epidemiology, providing a deeper understanding of the diet-disease nexus. This study contributes to the evolving field of personalized nutrition and emphasizes the potential of leveraging advanced computational tools to inform targeted nutritional interventions and public health strategies aimed at mitigating the prevalence of metabolic disorders in the Korean population.

Computational Investigation of Conformational Properties of Short Azapeptides: Insights from DFT Study and NBO Analysis

Azapeptides have gained much attention due to their ability to enhance the stability and bioavailability of peptide drugs. Their structural preferences, essential to understanding their function and potential application in the peptide drug design, remain largely unknown. In this work, we systematically investigated the conformational preferences of three azaamino acid residues in tripeptide models, Ac-azaXaa-Pro-NHMe [Xaa = Asn (4), Asp (5), Ala (6)], using the popular DFT functionals, B3LYP and B3LYP-D3. A solvation model density (SMD) was used to mimic the solvation effect on the conformational behaviors of azapeptides in water. During the calculation, we considered the impact of the amide bond in the azapeptide models on the conformational preferences of models 4-6. We analyzed the effect of the HB between the side-chain main chain and main-chain main-chain on the conformational behaviors of azapeptides 4-6. We found that the predicted lowest energy conformation for the three models differs depending on the calculation methods. In the gas phase, B3LYP functional indicates that the conformers tttANP-1 and tttADP-1 of azapeptides 4 and 5 correspond to the type I of β-turn, the lowest energy conformation with all-trans amide bonds. Considering the dispersion correction, B3LYP-D3 functional predicts the conformers tctANP-2 and tctADP-3 of azapeptide 4 and 5, which contain the cis amide bond preceding the Pro residue, as the lowest energy conformation in the gas phase. The results imply that azaAsx and Pro residues may involve cis-trans isomerization in the gas phase. In water, the predicted lowest energy conformer of azapeptides 4 and 5 differs from the gas phase results and depends on the calculational method. For azapeptide 6, regardless of calculation methods and phases, tttAAP-1 (β-I turn) is predicted as the lowest energy conformer. The results imply that the effect of the side chain that can form HBs on the conformational preferences of azapeptides 4 and 5 may not be negligible. We compared the theoretical results of azaXaa-Pro models with those of Pro-azaXaa models, showing that incorporating azaamino acid residue in peptides at different positions can significantly impact the folding patterns and stability of azapeptides.

Relationship between indoor noise perception and remote work during the COVID-19 pandemic

The coronavirus disease 2019 (COVID-19) pandemic has had a major influence on working patterns worldwide, given the various lockdown periods and the shift to remote working. As people's noise perception is known to be closely linked with their work performance and job satisfaction, investigating the noise perception in indoor spaces, especially in situations where people work from home, is crucial; however, studies on this aspect are limited. Thus, here, this study aimed to investigate the relationship between indoor noise perception and remote work during the pandemic. The study assessed how people who worked from home perceived indoor noise, and how it related with their work performance and job satisfaction. A social survey was conducted with respondents who worked from home during the pandemic in South Korea. A total of 1,093 valid responses were used for data analysis. Structural equation modeling was used as a multivariate data analysis method to simultaneously estimate multiple and interrelated relationships. The results showed that indoor noise disturbance significantly affected annoyance and work performance. Annoyance with indoor noise affected job satisfaction. Job satisfaction was found to have a significant impact on work performance, particularly on two dimensions of the work performance that are crucial for achieving organizations' goals. Moreover, one dimension of the work performance had a significant impact on annoyance. The study proposed that reducing negative perception of indoor noise and improvement of job satisfaction can lead to the maximization of one's work performance when working from home.