Browse Articles

Evaluating the performance of GPT-3.5, GPT-4, and GPT-4o in the Chinese National Medical Licensing Examination

Abstract This study aims to compare and evaluate the performance of GPT-3.5, GPT-4, and GPT-4o in the 2020 and 2021 Chinese National Medical Licensing Examination (NMLE), exploring their potential value in medical education and clinical applications. Six hundred original test questions from the 2020 and 2021 NMLE (covering five types of questions) were selected and input into GPT-3.5, GPT-4, and GPT-4o for response. The accuracy of the models across different question types and units was recorded and analyzed. Statistical methods were employed to compare the performance differences among the three models. GPT-4o demonstrated significantly higher overall accuracy than GPT-4 and GPT-3.5 (P < 0.001). In the 2020 and 2021 exams, GPT-4o achieved accuracy rates of 84.2% and 88.2%, respectively, with the highest accuracy observed in questions related to the digestive system (Unit 3), reaching 94.75%. GPT-4 showed moderate performance, while GPT − 3.5 had the lowest accuracy. Additionally, GPT-4o exhibited a clear advantage in complex question formats, such as case analysis questions (A3/A4 type) and standard matching questions (B1 type). GPT-4o outperformed its predecessors in the NMLE, demonstrating exceptional comprehension and problem-solving abilities in non-English medical examinations. This study provides important insights into the application and promotion of generative AI in medical education and clinical practice.

Semantic priming supports infants’ ability to learn names of unseen objects

Human language permits us to call to mind representations of objects, events, and ideas that we cannot witness directly, enabling us to learn about the world far beyond our immediate surroundings. When and how does this capacity emerge? To address this question, we evaluated infants at 12 and 15 months, asking whether they establish a representation of a novel noun’s meaning in the absence of any visible referents, and use this representation to identify a candidate referent when it later becomes available. During training, infants (67 12-month-olds; 67 15-month-olds) were primed with words and images of objects from a particular semantic neighborhood (e.g., fruits) and were also introduced to a novel noun (e.g., “a modi”), used to name a hidden object. During test, infants heard that noun again, this time with two unfamiliar objects present—one from the primed neighborhood (e.g., a dragon fruit) and the other from an unrelated semantic neighborhood (e.g., an ottoman). If infants can represent something about the meaning of the novel noun in the absence of a visible referent and then use such a representation when a candidate referent appears, then at test, they should prefer the object from the primed semantic neighborhood. At 15 months, infants succeeded. In contrast, 12-month-olds did not succeed on this task even after a full week of vocabulary training designed to boost the effect of priming. It is possible then that 12-month-olds’ representations of novel nouns’ meaning are not yet sufficiently rich (if any at all) to guide their choice of referent when one does appear. Together, these findings suggest that the capacity to establish a representation of a novel noun’s meaning in the absence of any visible referent and use this representation later to identify a candidate referent object emerges between 12 and 15 months.

Angiotensin converting enzyme 2 in patients with sepsis associate with comorbidities but neither with mortality nor with organ failure

Abstract High levels of circulating angiotensin converting enzyme 2 (ACE2) are associated with several chronic diseases and mortality risk. Less is known about the prognostic value of ACE2 in patients with sepsis. In the present study, we aimed to investigate the association between plasma ACE2 levels on admission to the ED and 28-day mortality, organ failure, and level of care in a prospectively recruited observational study sample. Six hundred patients with sepsis admitted to the emergency in Malmö 2013–2015 were included in the analysis. Uni- and multivariable binary logistic regression was conducted to investigate the association between ACE2 and 28-day mortality, organ failure, and level of care. Plasma ACE2 levels were increased in patients with male sex, high age, and comorbidities, including diabetes, cardiovascular disease, and cancer. Plasma ACE2 levels associated with hospitalization and intermediate care unit stay in univariate but not multivariate analysis. Plasma ACE2 did neither associate with 28-day mortality, OR 1.19 (95% CI 0.86–1.65, p = 0.29), nor with organ failure, OR 1.08 (95% CI 0.73–1.56. p = 0.72). Future studies investigating the dynamics of circulating ACE2 levels in patients with sepsis longitudinally are warranted.

Preventing the preventable: Assessing the burden of incessant caesarean deliveries in select Indian states using NFHS-5

The World Health Organization (WHO) recommends that life-saving Caesarean sections (CS) should account for 10–15% of deliveries. Southern states of India have good obstetric facilities and better public health systems than other parts of India. However, CS rates in select states are considerably higher. This study examines the prevalence and determinants of preventable CS among mothers in select Indian states, specifically the states that have high institutional deliveries (>95%) viz. Andhra Pradesh, Kerala, Karnataka, and Tamil Nadu. It further compares the complications faced by CS and vaginal delivery cases. Data has been derived from a cross-sectional survey, the National Family Health Survey-5. Bivariate and Logistic regression analyses were used. The main outcome measured is Preventable C-sections, defined as per conditions in Robson’s criteria of 10-group classification based on obstetric characteristics for a woman. Results show that the preventable CS rates in these selected states were much higher than at the national level. Key correlates were higher education and middle socioeconomic status. The study highlights the need for policy reforms, maternal education, and healthcare system improvements to reduce unnecessary CS in select states of India, where the health system is exceptionally good but , have another health burden in form of demand for unnecessary caesarean deliveries.

Publisher Correction: IVC treatment between primary and second TURBT may improve the prognosis of high-risk NMIBC patients receiving BCG treatment

Megafauna mobility: Assessing the foraging range of an extinct macropodid from central eastern Queensland, Australia

Understanding the factors that influence the geographic range of extinct megafaunal species is crucial for reconstructing their ecology and extinction dynamics. For extant herbivores, it has been demonstrated that large body mass provides the potential for greater geographic range. Allometric scaling relationships are observed in placental mammals but have not been well-established for marsupials, in particular, extinct marsupial megafauna. Here, we employ a phylogenetic generalised least squares regression model using extant macropodids to estimate home ranges for individuals from the extinct genus Protemnodon. The regression model predicts a mean home range of 11.6 ±  5.8 km2 This prediction, centred on Mt Etna caves, incorporates several distinct geological features with variable, known 87Sr/86Sr isotope ratios. Fossil Protemnodon individuals recovered from cave deposits at Mt Etna returned 87Sr/86Sr values similar to that of the host limestone, in which the cave systems formed, and the broader Mount Alma Formation. This similarity suggests that individuals foraged close to where they were fossilised, indicating a smaller home range than predicted. Smaller home ranges for individuals with a large body-mass were unexpected, attributed to a unique combination of individual behaviour, diet and/or locomotion regime within stable rainforest environments. Our results suggest that, foraging ranges in marsupial megaherbivores may be more strongly associated with environmental quality rather than body mass. New in-situ uranium-thorium and single-grain TT-OSL ages refine, and are in agreement with, previous interpretations of chronology, indicating that rainforest-adapted fauna persisted at Mt Etna until at least 280 ka. We propose that small home ranges in a stable environment, such as rainforests, predisposed these megafauna macropodids to extinction after 280ka, driven by an increasingly dry and unstable climate. Our results underscore the need for regionally specific biologies of individuals, populations and species when considering extinction pathways for Pleistocene fauna.

A combined model method was used to identify the main influencing factors of soil heavy metal pollution sources in Qian river, China

Does digital literacy affect farmers’ adoption of agricultural social services? An empirical study based on China Land Economic Survey data

Agricultural social services (ASS) are crucial in alleviating resource constraints and advancing agricultural modernization. Using the data derived from the China Land Economic Survey (CLES2022), this study empirically investigates how digital literacy influences farmers’ adoption of ASS, employing both Probit and Propensity score matching (PSM) models. Additionally, it explores the mediating roles of long-term production vision and part-time employment degrees in this relationship. The findings are: (1) digital literacy exhibited a statistically significant positive effect on farmers’ adoption of ASS at the 1% significant level. Moreover, this impact varied among participants in technology training, different education levels, and varying levels of risk preference; (2) long-term production vision and part-time employment degrees act as mediators, enhancing the positive impact of digital literacy on farmers’ adoption of ASS. Based on these findings, recommendations have been developed to improve farmers’ digital literacy, promote the adoption of ASS, and enhance farmers’ long-term production vision as well as degree of part-time.

Large field-of-view event-mode camera for high-precision epithermal neutron resonance imaging

Abstract A large-area event-mode camera system coupled with a $$^{6}$$ LiF-ZnS:Ag scintillator is applied for neutron resonance imaging (NRI) on the energy-resolved neutron imaging (ERNI) flight path, also known as Flight Path 5 (FP5), at the Los Alamos Neutron Science Center (LANSCE). This novel neutron imaging system, featuring a 120 $$\times$$ 120 mm $$^2$$ field of view, efficiently captures resonance information across the entire image in a single acquisition, significantly reducing beam time requirements compared to conventional energy-resolved neutron imaging systems. High-quality neutron radiographs with enhanced spatial resolution are achieved through the reconstruction of neutron events based on observations of individual photons emitted from the scintillator. The system demonstrates reduced background through neutron/gamma discrimination capabilities while maintaining sharpness across a large fields of view. In the measurements presented here, a spatial resolution of approximately 340 $$\mu$$ m was achieved using center-of-gravity photon cluster centroiding. We demonstrate the system’s capability for quantitatively determining isotopic distributions in various thin samples, as well as automatically reconstructing complex scenes with overlapping resonances from diverse samples. These results are obtained using standard data analysis tools, despite the relatively slow $$^{6}$$ LiF-ZnS:Ag scintillator, which may not be optimal for absorption resonance detection. The capabilities demonstrated here offer a valuable, versatile, and cost-effective solution for high spatial and temporal resolution, large field-of-view energy-resolved neutron imaging, with potential applications across various scientific and industrial domains.

Evaluation of intake of aged garlic extract and organosulfur compounds on progressive hearing loss in DBA/2J mice

Garlic is rich in organosulfur compounds with high antioxidant capacity and is known to have various health benefits. Aged garlic is a particularly effective source of these active compounds because it contains fewer toxic components. This study evaluated the effects of S-allyl cysteine (SAC), a major functional organosulfur compound in garlic, and aged garlic extract (AGE) on the suppression of progressive hearing loss. SAC and AGE were dissolved in drinking water at 1% (w/w) and administered to DBA/2J mice, a model of early progressive hearing loss, for 12 weeks, starting at 4 weeks of age. While the results revealed a trend toward weight loss in the SAC group, the weight of the AGE group was comparable to that of the control group, and no adverse events were observed in either group. The hearing ability of the mice was measured using auditory brainstem responses at 4, 8, 12, and 16 weeks of age. Hearing loss at 8 and 16 kHz progressed over the 12-week period, with neither sample inhibiting hearing loss. In contrast, another organosulfur compound, N-acetylcysteine (NAC), administered in 1% w/w drinking water and evaluated for hearing loss over time, significantly suppressed hearing loss progression in DBA/2J mice. These results indicate that the NAC and SAC differ in their ability to prevent hearing loss and demonstrate that the inhibitory effects of functional food components on hearing loss can be evaluated over a short period in DBA/2J mice.

PoulTrans: a transformer-based model for accurate poultry condition assessment

Abstract Recent advances in deep learning have significantly enhanced the accuracy of poultry image recognition, particularly in assessing poultry conditions. However, developing intuitive decision support tools remain a significant challenge. To address this, we present PoulTrans, an innovative image captioning framework that leverages a Convolutional Neural Network (CNN) integrated with a CSA_Encoder-Transformer architecture to generate detailed poultry status reports. This model incorporates visual features extracted by CNNs into the Channel Spatial Attention Segmentation Encoder (CSA_Encoder), which produces segmented channel and spatial attention outputs. To optimize multi-level attention and improve the semantic precision of the status descriptions, we introduced a Channel Spatial Memory-Guided Transformer (CSMT) and a novel PS-Loss function. The performance of PoulTrans was tested on the PSC-Captions dataset, achieving top scores of 0.501, 0.803, 4.927, 0.608, and 1.882 for the BLEU-4, ROUGE-L, CIDEr, SPICE, and Sm metrics, respectively. Comprehensive analyses and experiments have validated the effectiveness and reliability of our model, providing advanced tools for automated poultry status generation and enhancing the digital experience for poultry farmers. Our code is available at: https://github.com/kong1107800/PoulTrans.

From vulnerability to duplicity: Examining the connection between childhood adversity and deception

Deception research has traditionally evaluated how individual differences like personality traits and demographics correlate with lying. However, the establishment of adverse childhood experiences (ACEs) as an individual difference that also links to deception remains underexplored. To this end, the present study (N = 784 students) investigated the relationship between ACEs and deception in adulthood. Results indicated that individuals with more (versus less) adverse childhood experiences, particularly those involving maltreatment and victimization, reported more daily white and big lies, independent of aversive personality traits like narcissism and Machiavellianism. Consistent with other studies on individual differences and deception, the effect sizes were small, but systematic. Together, these findings support the dispositional honesty hypothesis, indicating that foundational childhood experiences and events can shape or signal deceptive behavior. Generally, the study contributes to our underexamined knowledge base of the developmental antecedents of lying, emphasizing the role that adversity plays during childhood to influence deceptive behavior beyond commonly studied personality traits.

CCNE2 promotes cisplatin resistance and affects prognosis of head and neck squamous cell carcinoma by targeting MNAT1

The role of university entrepreneurship support and student diversity in student entrepreneurial intention: a social identity theory perspective

One of the most significant shortcomings of the extant literature is the dearth of conceptualizations and empirical testing pertaining to entrepreneurial diversity and the underlying social identities. Accordingly, this research aims to empirically investigate the influence of entrepreneurial support at the university level on students’ entrepreneurial intention from a diversity and social identity perspective. The research utilizes fsQCA (fuzzy set Qualitative Content Analysis) and a set of observable and underlying diversity factors (gender, age, previous academic performance, and parental educational attainment) to elucidate the causal structures that influence the formation of entrepreneurial intention among university students. The findings illustrate the multifaceted nature of the entrepreneurial intention formation, with first-generation status being a necessary condition for low entrepreneurial intention. The findings underscore the importance of university entrepreneurial support in the formation of entrepreneurial intention especially for younger students and first-generation students. The study contributes to the literature on entrepreneurial diversity and social identities in entrepreneurship by identifying the underlying factors that contribute to high and low entrepreneurial intention among university students.

Selection and validation of stable reference genes in potato infected by Pectobacterium atrosepticum using real-time quantitative PCR

Optimizing China’s carbon quota allocation for peak emissions: A novel AMLC framework tailored to regional dynamics

As China advances its carbon reduction efforts, the equitable and efficient allocation of initial carbon quotas among provinces has become increasingly critical. This study proposes a novel two-stage Adjusting Measures to Local Conditions (AMLC) framework that integrates regional development disparities and marginal abatement costs into the quota allocation process. Employing a PSO-LSSVM model for carbon emission estimation and a hierarchical clustering algorithm, this study categorizes 30 Chinese provinces into six distinct types, facilitating a regionally adaptive allocation strategy. Key findings include: (1) Between 2003 and 2022, carbon emissions exhibited a structural shift from energy-abundant, carbon-exporting provinces with inefficient industrial structures to energy-deficient, carbon-importing provinces, intensifying regional disparities. (2) Equity-based allocation schemes tend to favor regions with abundant carbon sinks, yet all provinces in the sixth category experience quota deficits; conversely, efficiency-based allocation disproportionately benefits initially efficient regions, exacerbating interprovincial disparities. (3) The AMLC framework enables 17 provinces across all clusters to achieve quota surpluses while significantly enhancing national green productivity, demonstrating its superiority in promoting high-quality economic development. These findings provide valuable insights for policymakers in designing macroeconomic strategies that balance equity and efficiency in carbon quota distribution, thereby supporting China’s transition toward peak emissions and sustainable growth.

Artery segmentation and atherosclerotic plaque quantification using AI for murine whole slide images stained with oil red O

Abstract Atherosclerosis is the leading cause of death in Western industrial nations. To study the etiology of plaque progression, atherosclerotic mouse models are widely used. Traditionally, analyzing the obtained histological whole slide images of Oil Red O-stained aortic roots required manual segmentation. To accelerate this process, an artificial intelligence-driven solution is proposed that comprises three stages: (1) defining the region of interest (ROI) of the aortic root using a YOLOv8l object detector, (2) applying supervised machine learning with ensembles of U-Net++ networks for artery segmentation using ROI masks, and (3) performing plaque segmentation within arterial walls with the unsupervised W-Net method. To establish a robust segmentation pipeline, we benchmark our methods using manually created masks ( $$\text {n}=6085$$ for artery segmentation, $$\text {n}=1089$$ for plaque segmentation). A key finding is that an ensemble of U-Net++ networks applied on ROI masks outperformed single network architectures. Through a novel combination strategy, the ensemble output can be easily modified, paving the way for a quick and robust application in the lab. Finally, a case study utilizing published mouse data ( $$\text {n}=373$$ slices) underscored the ability of our optimized pipeline to replicate human-made plaque predictions with a high correlation (Pearson’s $$\text {r}=0.91$$ ) and reproduce biological insights derived from manual analysis.

Repeatability of wildlife surveys for estimating abundance: A method to assess the consistency of detection probability and animal availability

A primary consideration of abundance studies that use unmarked animals is whether survey counts accurately reflect the population size or if unknown variation in animal movement or detection probability biases counts irrespective of population size. We posited that high repeatability in counts among temporally replicated surveys would indicate that counts are a good index of abundance. We temporally replicated 49 nocturnal spotlight surveys of white-tailed deer (Odocoileus virginianus) up to three times each (n = 128 total samples) to test the repeatability of this commonly used wildlife monitoring technique. Repeatability was high (R = 0.86), suggesting spring spotlight surveys provide a reliable index of deer population size in Iowa, USA. Fourteen percent of the variation among replicated counts was explained by day of year and, to a lesser degree, a vegetation green-up index. Detection probability was high (~0.70) early in the sampling season and declined considerably during the following 6 weeks. Deer abundance was greater at sites with higher percent landcovers of forest and hay/pasture and was lower at sites with higher landcover in crops. Our findings suggest deer managers should sample prior to green-up in the spring to maximize the proportion of the population that is detectable, and that accounting for seasonality on detection estimation is important for reliable abundance estimates if sampling occurs over a range of phenological progression. Finally, we show that temporal replication of surveys is a logistically feasible method to assess the reliability of abundance estimates from study designs that are normally conducted with single visits.

Simultaneous production of fatty acids and amino polysaccharides from Norway spruce hydrolysates using oleaginous Mucor circinelloides

Abstract Lignocellulose is an abundant raw material and renewable carbon source for the production of single cell oils which can replace plant-derived oils in food, feed, fuels, and oleochemicals. Mucor circinelloides produces both fatty acids and amino polysaccharides, such as chitin and chitosan. This study evaluates hydrolysates of Norway spruce (Picea abies) as a carbon source for their simultaneous production. Cultivation in spruce hydrolysate media yielded 15.8 g/L of biomass, with fatty acids comprising ~ 50% of the cell dry weight and amino polysaccharides up to 8.5%. The fatty acid methyl ester (FAME) content and fatty acid profile were comparable to glucose fermentation. Optimal harvesting times ranged from 72 to 120 h, depending on desired yields. These findings demonstrate that Norway spruce hydrolysates are a viable and sustainable substrate for microbial lipid and polysaccharide production, supporting their potential use in biotechnology and industrial applications.

Vibration control of a class of flexible mechanical systems with output constraints based on partial differential equations

Vibration suppression in flexible mechanical systems (FMSs) significantly enhances the precision and stability of equipment, extending its operational life. This technology is extensively applied across various sectors, including aerospace, robotics, and precision manufacturing. This paper introduces a robust control scheme leveraging a Partial Differential Equations (PDEs) model and Barrier Lyapunov Function (BLF), applied through backstepping technology to manage a flexible mechanical system modeled as an Euler-Bernoulli beam with a centrally attached rigid body. The control laws we have formulated are designed to effectively dampen vibrations and rotations, thereby ensuring system stability despite the presence of environmental disturbances. Throughout the control process, the system output consistently stays within the predefined safety limits. Comparative simulations further validate the effectiveness of the proposed control strategy, showing that it can effectively counteract unforeseen disturbances while ensuring that the output remains within the specified constraints.