Search Results (6,508)

This study investigates how software developers discuss usability on Stack Overflow through an analysis of posts from 2008 to 2024. Despite recognizing the importance of usability for software success, there is a limited amount of research on developer engagement with usability topics. Using mixed methods that combine quantitative metric analysis and qualitative content review, we examine temporal trends, comparative engagement patterns across eight non-functional requirements, and programming context-specific usability issues. Our findings show a significant decrease in usability posts since 2010, contrasting with other non-functional requirements, such as performance and security. Despite this decline, usability posts exhibit high resolution efficiency, achieving the highest answer and acceptance rates among all topics, suggesting that the community is highly effective at resolving these specialized questions. We identify distinctive platform-specific usability concerns: web development prioritizes responsive layouts and form design; desktop applications emphasize keyboard navigation and complex controls; and mobile development focuses on touch interactions and screen constraints. These patterns indicate a transformation in the sharing of usability knowledge, reflecting the maturation of the field, its integration into frameworks, and the migration to specialized communities. This first longitudinal analysis of usability discussions on Stack Overflow provides insights into developer engagement with usability and highlights opportunities for integrating usability guidance into technical contexts. Full article

This paper assesses the correlates of perceived well-being among young women aged 18 to 30 in five Balkan cities: Athens, Greece; Plovdiv, Bulgaria; Bucharest, Romania; Nis, Serbia; and Shtip, North Macedonia, by integrating urban, travel behavioural, and socio-economic features. A cross-sectional survey was employed using standard questionnaires including the Warwick–Edinburgh Mental Well-being Scale (WEMWBS), the short version of the International Physical Activity Questionnaire (IPAQ), and the adapted ALPHA environmental questionnaire. To answer research questions, linear regression models were developed to analyse predictors of well-being at both regional and national levels. Results show that neighbourhood and mobility features play a significant role in shaping mental well-being. Access to walkable sidewalks, green spaces, mixed land-use structure, and attractive local facilities (e.g., shops, recreational centres in the neighbourhood) were consistently associated with higher levels of well-being. Conversely, perceived insecurity, especially at night or regarding bicycle theft, significantly reduced well-being. Physical activity levels, particularly days of walking and vigorous activity, showed strong positive associations, underscoring the role of active lifestyles in promoting mental health. Socio-economic variables, including financial status, relationship status, and work status, were also found to be linked to perceived well-being. Cycling-related variables may affect Greek well-being up to 16.5 times. Perception of crime during the night may negatively affect both Bulgarian and Serbian well-being (up to 10 times), while Romanian well-being is mostly affected by the existence of shopping facilities. Finally, the most impactful factors for well-being in North Macedonia refer to cycling safety and scooter accessibility. Full article

This paper presents a systematic evaluation of large language models (LLMs) and retrieval-augmented generation (RAG) approaches for question answering (QA) in the low-resource Kazakh language. We assess the performance of existing proprietary (GPT-4o, Gemini 2.5-flash) and open-source Kazakh-oriented models (KazLLM-8B, Sherkala-8B, Irbis-7B) across closed-book and RAG settings. Within a three-stage evaluation framework we benchmark retriever quality, examine LLM abilities such as knowledge-gap detection, external truth integration and context grounding, and measures gains from realistic end-to-end RAG pipelines. Our results show a clear pattern: proprietary models lead in closed-book QA, but RAG narrows the gap substantially. Under the Ideal RAG setting, KazLLM-8B improves from its closed-book baseline of 0.427 to reach answer correctness of 0.867, closely matching GPT-4o’s score of 0.869. In the end-to-end RAG setup, KazLLM-8B paired with Snowflake retriever achieved answer correctness up to 0.754, surpassing GPT-4o’s best score of 0.632. Despite improvements, RAG outcomes show an inconsistency: high retrieval metrics do not guarantee high QA system accuracy. The findings highlight the importance of retrievers and context grounding strategies in enabling open-source Kazakh models to deliver competitive QA performance in a low-resource setting. Full article

This paper presents the first rigorous empirical investigation into a fundamental question of cryptocurrency valuation: Are cryptocurrency prices in line with the prices of fundamental assets? To answer this, we analyze the nine largest cryptocurrencies by market capitalization—Bitcoin (BTC), Ethereum (ETH), Solana (SOL), Binance Coin (BNB), Ripple (XRP), Cardano (ADA), Litecoin (LTC), Tron (TRX), and the stablecoin DAI—against a suite of traditional benchmarks, including major fiat currencies (EUR, CAD, JPY), gold, and the S&P500 index. Our dataset spans from 1 January 2014 to 30 June 2025, with start dates varying for newer cryptocurrencies to ensure robust time series analysis. Guided by the asset pricing theory, we formulate a martingale test: if a cryptocurrency is priced in line with a fundamental numeraire asset, its price ratio relative to that numeraire must follow a martingale process. Our extensive empirical analysis reveals that the prices of major cryptocurrencies (BTC, ETH, SOL, BNB) consistently reject the martingale hypothesis when traditional assets (currencies, gold, equities) serve as the numeraire, indicating a decoupling from fundamental valuation anchors. Conversely, when Bitcoin or Ethereum itself is used as the numeraire, most smaller cryptocurrencies are priced in line with these crypto benchmarks, suggesting an internal valuation ecosystem that operates independently of traditional finance. Full article

This paper outlines a view of religious language that revolves around the notion of informed interpretation. The view can be summed up by saying that there is no fact of the matter independently of context and informed interpretation as to whether some religious statement or expression has cognitive content, or what that content may be. Where informed interpretation of religious language is impossible, we can give no answer to the question of what the content of a given statement or expression is. Equally, there can be no answer to the question of what that statement or expression presupposes or implies. If this is correct, then the idea that there can be a general and abstract philosophical analysis or theory of religious language should be called into question. Full article

Contractual agreements often contain clauses that are unfair, creating unjust suffering in one party to the agreement. ContractNerd leverages advanced Large Language Models (LLMs) to analyze contractual agreements and identify issues across four categories: missing clauses, unenforceable clauses, legally sound clauses, and legal but risky clauses. By using a structured methodology that integrates LLM-based clause comparison, enforceability checks against jurisdiction-specific regulations, and assessments of risk-inducing traits, ContractNerd provides a comprehensive analysis of contractual terms. To evaluate the tool’s effectiveness, we compare its analyses with those from existing platforms on rental clauses that have led to court litigation. ContractNerd’s interface helps users (both drafters and signing parties) to navigate complex contracts, offering actionable insights to flag legal risks and disputes. Full article

Large Language Models (LLMs) represent a major step in AI development and are increasingly used in daily applications. However, they are prone to hallucinations, generating claims that contradict established facts, deviating from prompts, and producing inconsistent responses when the same prompt is presented multiple times. Addressing these issues is challenging due to the lack of comprehensive and easily assessable benchmark datasets. Most existing datasets are limited in scale and scope and rely on multiple-choice questions, which are insufficient for evaluating the generative capabilities of LLMs. To assess hallucination in LLMs, this paper introduces a comprehensive benchmark dataset consisting of over 20,000 unique prompts (more than 75,000 prompts in total) across eight domains. These prompts are designed to elicit definitive, concise, and informative answers. The dataset is divided into two segments: one publicly available for testing and assessing LLM performance, and a hidden segment for benchmarking various LLMs. In our experiments, we tested nine State-of-The-Art (SoTA) models, GPT-4o, GPT-3.5, LLama 2 7B, LLama 3 8B, Gemini 1.0 Pro, Mixtral 8x7B, Zephyr 7B, Deepseek-r1-7b, and Qwen2.5-14B, revealing that overall factual hallucination ranges from 48% to 82% on the public dataset and 31% to 76% on the hidden benchmark. Prompt Misalignment Hallucination ranges up to 95% in the public dataset and up to 94% in the hidden counterpart. Average consistency ranges from 21% to 61% and 44% to 63%, respectively. Domain-wise analysis reveals that LLM performance significantly deteriorates when asked for specific numeric information, whereas it performs moderately with queries involving persons, locations, and dates. Our dataset demonstrates its efficacy and serves as a comprehensive benchmark for evaluating LLM performance. Full article

Various clinical studies aimed at modifying the progression of idiopathic Parkinson’s disease have been unsuccessful. Similarly, several nutritional trials using bioactive compounds have shown positive effects for patients but have also failed to slow or reduce the disease’s progression. This repeated failure is likely because these studies ignore the extremely slow neurodegenerative process, which unfolds over many years. The molecular mechanism behind the loss of neuromelanin-containing dopaminergic neurons in the nigrostriatal system in idiopathic Parkinson’s disease remains unclear. This is a conceptual/theoretical review based mainly on mechanistic and preclinical evidence, with no direct clinical data. However, research suggests that aminochrome, an endogenous neurotoxin, may trigger the degeneration of these neurons through a single-neuron degeneration model. In this model, aminochrome selectively destroys individual neurons without spreading to neighboring cells. Aminochrome is produced during neuromelanin synthesis, a process that is normally harmless because protective enzymes like DT-diaphorase and glutathione transferase M2-2 neutralize aminochrome’s neurotoxic effects. Increasing the levels of these enzymes could offer neuroprotection. The KEAP1/NRF2 signaling pathway is critical for regulating antioxidant enzymes, such as DT-diaphorase and glutathione transferase M2-2. Importantly, specific bioactive compounds from food can activate this pathway, increasing the production of these protective enzymes. For instance, the omega-3 fatty acids eicosapentaenoic acid (EPA) and docosahexaenoic acid (DHA), along with astaxanthin—a compound present in cold-water fish like salmon—have been demonstrated to enhance enzyme expression. This connection leads to a compelling question: Could dietary interventions help prevent idiopathic Parkinson’s disease? Answering this will require further research. Full article

Functional magnetic resonance imaging (fMRI) is commonly used to investigate the neural bases of aging and psychological disorders. However, the BOLD signal captured by fMRI is affected by many factors that are non-neural in origin. We tested how vascular health risks, which often go unmeasured in neuroimaging studies, and aging interact to modify the shape and/or timing of the HRF, which then affect the differences in patterns of brain activity in a task-evoked memory encoding paradigm. Adult participants (aged 20–74) answered questions about their health and underwent two fMRI tasks: viewing a flashing checkerboard and a memory encoding task. Aging and vascular risk had the largest impacts on the maximum peak value of the HRF. Using a subject-specific HRF resulted in a dampening of brain activity in task-positive and task-negative regions. Across three vascular risk factors, using a subject-specific HRF resulted in more consistent brain regions that reached significance and larger effect sizes compared with the canonical HRF. These findings serve as a cautious tail when interpreting task-evoked fMRI activity, especially in populations experiencing alterations to brain vasculature including many older adults and people with neurocognitive disorders like Alzheimer’s disease and related dementias. Full article

Background: Direct clinical training on real patients is essential in dental education. However, the declining patient inflow increasingly challenges this objective. This cross-sectional study aimed to assess patients’ experiences and preferences to derive recommendations for improving patient recruitment. Material and Methods: Over a period of one year, patients treated by students in the courses and final examinations at the dental school of conservative dentistry were questioned using a specially designed questionnaire and reviewed using their medical records. They were asked about their complete treatment process, and patient files were used to record socio-demographic as well as economic and appointment-specific data. Results: We analysed 297 patients (142 women, 47.8%; 155 men, 52.2%) treated by undergraduates across two semesters (four courses) and two final examinations. Median age was 57.0 years (IQR 46–67; mean 55.2, SD 15.2; range 14–85) with no sex-based difference (p > 0.05). Arrival was predominantly by car (72.7%, n = 216); median one-way distance was 20.5 km (IQR 11.2–32.1); and 58.4% were employed, while 41.6% were not employed (33.7% retired, 7.9% unemployed). The leading reason for course attendance was “satisfaction with previous treatments” (65.32%). Information sources were reported by 290/297 (98%); the most common was already being a course patient (143, 48.1%). Most patients attended one appointment (109, 36.7%). Median travel cost per appointment (including parking) was €17.0 (typically €10.0–€23.5). Of 285 respondents, 93.68% answered “Yes” to satisfaction with student treatment. Conclusions: Important steps include enhancing parking facilities, optimizing recall systems and appointment accessibility, and strengthening relationships with regular patients to encourage word-of-mouth referrals. The main focus is to maintain high clinical quality, ensure affordability, and further reduce patient copayments where possible. Full article

Background/Objectives: Artificial intelligence (AI) is gaining importance in clinical pharmacology, supporting therapeutic decisions and the prediction of drug interactions, although its applications have significant limitations. The aim of the study was to evaluate the accuracy of the responses of four large language models (LLMs), namely ChatGPT-4o, ChatGPT-3.5, Gemini Advanced 2.0, and DeepSeek, in the field of clinical pharmacology and drug interactions, as well as to analyze the impact of prompting and questions from the National Specialization Examination for Pharmacists (PESF) on the results. Methods: In the analysis, three datasets were used: 20 case reports of successful pharmacotherapy, 20 reports of drug–drug interactions, and 240 test questions from the PESF (spring 2018 and autumn 2019 sessions). The responses generated by the models were compared with source data and the official examination key and were independently evaluated by clinical-pharmacotherapy experts. Additionally, the impact of prompting techniques was analyzed by expanding the content of the queries with detailed clinical and organizational elements to assess their influence on the accuracy of the obtained recommendations. Results: The analysis revealed differences in the accuracy of responses between the examined AI tools (p < 0.001), with ChatGPT-4o achieving the highest effectiveness and Gemini Advanced 2.0 the lowest. Responses generated by Gemini were more often imprecise and less consistent, which was reflected in their significantly lower level of substantive accuracy (p < 0.001). The analysis of more precisely formulated questions demonstrated a significant main effect of the AI tool (p < 0.001), with Gemini Advanced 2.0 performing significantly worse than all other models (p < 0.001). An additional analysis comparing responses to simple and extended questions, which incorporated additional clinical factors and the mode of source presentation, did not reveal significant differences either between AI tools or within individual models (p = 0.34). In the area of drug interactions, it was also shown that ChatGPT-4o achieved a higher level of response accuracy compared with the other tools (p < 0.001). Regarding the PESF exam questions, all models achieved similar results, ranging between 83 and 86% correct answers, and the differences between them were not statistically significant (p = 0.67). Conclusions: AI models demonstrate potential in the analysis of clinical pharmacology; however, their limitations require further refinement and cautious application in practice. Full article

Urban agglomerations are key to balancing carbon emissions (CEs) and ecosystem services (ESs), yet structural imbalances exist between LUCE and ESs due to the lack of standardized frameworks and clear governance strategies. This study investigates the relationship between LUCE and ESs in the Chang-Zhu-Tan urban agglomeration using multi-source data from 2010 to 2023. The study aims to address three main research questions: (1) How do LUCE and ES networks evolve over time? (2) What factors drive their heterogeneity? (3) How do urbanization and ecological restoration impact LUCE and ES network dynamics? To answer these, we apply centrality metrics and develop heterogeneity indices to evaluate connectivity, accessibility, and driving factors. The findings show that both LUCE and ES networks exhibit corridor-like structures, with asymmetric node distributions. The LUCE-Network’s degree centrality increased from 0.16 to 0.29, while the ES-Network’s rose from 0.16 to 0.23. Heterogeneity was initially positive but turned negative by 2023, indicating a shift from LUCE dominance to an increased emphasis on ES. This transition was influenced by urbanization, land use changes, and ecological restoration efforts. Notably, the proportion of built-up land (X11) grew from 0.0187 in 2010 to 0.1500 in 2023, intensifying the disparity between LUCE and ESs. Similarly, urbanization (X7) surged to 0.1558 in 2023, increasing CEs and the demand for ESs. A collaborative pathway is proposed to address these challenges, involving controlled urban development, restoration of green spaces, and prioritizing multimodal transport and energy efficiency. This framework offers actionable diagnostics for improving low-carbon and ecological governance in urban agglomerations. Full article

Background: Heart failure (HF) affects over 64 million people globally, imposing substantial morbidity, mortality, and economic burdens. Despite advances in guideline-directed therapies, adherence remains suboptimal due to low health literacy and complex regimens. ChatGPT, an advanced large language model by OpenAI, offers conversational capabilities that could enhance HF education, management, and research. This systematic review synthesizes evidence on ChatGPT’s applications in HF, evaluating its accuracy in patient education and question-answering, enhancing readability, and clinical documentation/symptom extraction. Methods: Following PRISMA guidelines, we searched PubMed, Embase, and Cochrane up to July 2025 using the terms “ChatGPT” and “heart failure”. Inclusion: Studies on ChatGPT (3.5 or 4) in HF contexts, such as in education, readability and symptom extraction. Exclusion: Non-HF or non-ChatGPT AI. Data extraction covered design, objectives, methods, and outcomes. Thematic synthesis was applied. Results: From 59 records, 7 observational studies were included. Themes included patient education/question-answering (n = 5), readability enhancement (n = 2), and clinical documentation/symptom extraction (n = 1). Accuracy ranged 78–98%, with high reproducibility; readability improved to 6th–7th grade levels; and symptom extraction achieved up to 95% F1 score, outperforming traditional machine learning baselines. Conclusions: ChatGPT shows promise in HF care but requires further randomized validation for outcomes and bias mitigation. Full article

We investigate two closely related Lattice Gas Cellular Automata models of the interplay of aggregation of biological cells and synchronization of intracellular oscillations (“clocks”): clock-dependent aggregation, where the adhesive forces between cells depend on their relative clock phases (akin to so-called “swarmalators”), and simple adhesive aggregation, where they do not. Patterns of aggregation are similar for comparable ranges of parameters. However, while simple adhesive aggregation is quite similar to perikinetic aggregation, we show that clock-dependent aggregation differs in subtle ways. We found that it tends to inhibit coalescence of patterns and regularizes aggregate shapes, and, unintuitively, tends to enhance overall synchronization of clocks. Specifically, clock-dependent aggregation showed higher average circularity of aggregates and a larger value of Kuramoto’s r, measuring synchrony. Our results add to the growing literature on swarmalator models and give additional theoretical backing to the previously proposed idea that intracellular oscillatory processes may serve to regularize pattern formation, e.g., in chondrogenic condensation in embryonic chicken limbs. They thus contribute to a partial answer to the question: In the feedback between clocks and attraction in swarmalator models, how important is the effect of clocks on attraction? The detailed, systematic comparison of the results of these two types of aggregation is novel. Full article

This review investigates and analyzes the state of the art on scientific evidence related to educational interventions to improve air quality indoors and outdoors through a mapping review. The review followed proposed guidelines for mapping reviews in environmental sciences and the steps described in the Template for a Mapping Study Protocol. The search was conducted in PubMed, Web of Science, Embase, Cinahl, and Google Scholar with no language restrictions, and was completed in January 2025. Three filters were applied: search, selection with inclusion and exclusion criteria (PECOS strategy), and data extraction. Two independent reviewers assessed article eligibility, and disagreements were resolved by a third researcher. Twenty-four studies that met the eligibility criteria were included. Five research questions were answered. Studies published between 1977 and 2024 were included, totaling 7289 participants aged 12 to 85. The geographic distribution was concentrated in China (five studies) and the United States (four studies), followed by South Korea, India, Australia, and other countries, with fewer publications. The methodological predominance was experimental studies; observational studies were also analyzed, although less frequently. The period with the greatest increase in the number of publications was between 2020 and 2024. The educational methods most commonly used in the studies were lectures and the delivery of information leaflets. Particulate matter with diameters of 2.5 μm and 10 μm (PM_2.5 and PM₁₀) were the most widely investigated pollutants in the studies. From our analyses, it was observed that the educational interventions to improve air quality, adopted in the selected studies, resulted in the acquisition of knowledge about the environmental effects and the importance of individual actions. The changes in behavior included the adoption of more sustainable practices and an improvement in air quality in the environment, with a significant reduction in pollutant emissions. We conclude that interventions through environmental education demonstrate great potential to improve air quality. Based on the mapped evidence, governments and global policymakers can use this information to develop new strategies or improve existing ones to reduce air pollution in affected environments and regions. Full article