Language bias stands as a noteworthy concern in visual question answering (VQA), wherein models tend to rely on spurious correlations between questions and answers for prediction. This prevents the models from effectively generalizing, leading to a decrease in performance. In order to address this bias, we propose a novel modality fusion collaborative de-biasing algorithm (CoD). In our approach, bias is considered as the model\'s neglect of information from a particular modality during prediction. We employ a collaborative training approach to facilitate mutual modeling between different modalities, achieving efficient feature fusion and enabling the model to fully leverage multimodal knowledge for prediction. Our experiments on various datasets, including VQA-CP v2, VQA v2, and VQA-VS, using different validation strategies, demonstrate the effectiveness of our approach. Notably, employing a basic baseline model resulted in an accuracy of 60.14% on VQA-CP v2.

[This corrects the article DOI: 10.3389/fonc.2023.1239932.].

UNASSIGNED: The demand for healthcare is increasing globally, with notable disparities in access to resources, especially in Asia, Africa, and Latin America. The rapid development of Artificial Intelligence (AI) technologies, such as OpenAI\'s ChatGPT, has shown promise in revolutionizing healthcare. However, potential challenges, including the need for specialized medical training, privacy concerns, and language bias, require attention.
UNASSIGNED: To assess the applicability and limitations of ChatGPT in Chinese and English settings, we designed an experiment evaluating its performance in the 2022 National Medical Licensing Examination (NMLE) in China. For a standardized evaluation, we used the comprehensive written part of the NMLE, translated into English by a bilingual expert. All questions were input into ChatGPT, which provided answers and reasons for choosing them. Responses were evaluated for \"information quality\" using the Likert scale.
UNASSIGNED: ChatGPT demonstrated a correct response rate of 81.25% for Chinese and 86.25% for English questions. Logistic regression analysis showed that neither the difficulty nor the subject matter of the questions was a significant factor in AI errors. The Brier Scores, indicating predictive accuracy, were 0.19 for Chinese and 0.14 for English, indicating good predictive performance. The average quality score for English responses was excellent (4.43 point), slightly higher than for Chinese (4.34 point).
UNASSIGNED: While AI language models like ChatGPT show promise for global healthcare, language bias is a key challenge. Ensuring that such technologies are robustly trained and sensitive to multiple languages and cultures is vital. Further research into AI\'s role in healthcare, particularly in areas with limited resources, is warranted.

暂无摘要。

The rainfrogs of the genus Pristimantis are one of the most diverse groups of vertebrates, with outstanding reproductive modes and strategies driving their success in colonizing new habitats. The rate of Pristimantis species discovered annually has increased continuously during the last 50 years, establishing the remarkable diversity found in this genus. In this paper the specifics of publications describing new species in the group are examined, including authorship, author gender, year, language, journal, scientific collections, and other details. Detailed information on the descriptions of 591 species of Pristimantis published to date (June 2022) were analyzed and extracted. John D. Lynch and William E. Duellman are the most prolific authors, yet Latin American researchers have scaled up and continued the description processes since the 1990s. The most common language used for descriptions is English, followed by Spanish. The great majority of authors have described only one species. The largest proportion of authors who have participated in the descriptions is of Ecuadorian nationality. Ecuador is the country with the highest description rate per year (3.9% growth rate). Only 20% of the contributions have included women and only 2% have featured women as principal authors. 36.8% of the species described are in the Not Evaluated or Data Deficient categories under the IUCN global red list. The importance of enhancing the descriptions in Spanish is emphasized and the inclusion based on equal access to opportunities for female researchers in Pristimantis taxonomy is encouraged. In general, if the current trends in Pristimantis descriptions continue, in ten years, a total of 770 or more species described could be expected.
ResumenLas ranas de la lluvia del género Pristimantis es uno de los grupos de vertebrados más diversos, con una variedad de modos reproductivos y estrategias que impulsan su éxito en la colonización de nuevos hábitats. La tasa de especies de Pristimantis descubiertas anualmente ha aumentado continuamente durante los últimos 50 años, estableciendo la notable diversidad encontrada en este género. En este artículo, examinamos los detalles de las publicaciones que describen nuevas especies en el grupo, incluida la autoría, el año, el idioma, la revista, el género, las colecciones científicas y otros detalles. Analizamos y extrajimos información detallada sobre las descripciones de 591 especies de Pristimantis publicadas hasta la fecha (junio 2022). John D. Lynch y William E. Duellman son los autores más prolíficos, pero los investigadores latinoamericanos han ampliado y continuado los procesos de descripción desde la década de 1990. El idioma más común utilizado para las descripciones es el inglés, seguido del español. La gran mayoría de los autores han descrito una sola especie. La mayor proporción de autores que han participado en las descripciones es de nacionalidad ecuatoriana. Ecuador es el país con la tasa de descripción más alta por año (tasa de crecimiento del 3,9%). Solo el 20% de las contribuciones han incluido a mujeres y solo el 2% las ha presentado como autoras principales. El 36,8% de las especies descritas se encuentran en las categorías No evaluadas o Datos insuficientes de la lista roja mundial de la UICN. Destacamos la importancia de potenciar las descripciones en español y fomentar la inclusión de mujeres investigadoras en la taxonomía de Pristimantis. En general, si continúan las tendencias actuales en las descripciones de Pristimantis, en 10 años se podría esperar un total de 770 o más especies descritas.

暂无摘要。

To assess the impact of restricting systematic reviews of conventional or alternative medical treatments or diagnostic tests to English-language publications.
We systematically searched MEDLINE (Ovid), the Science Citation Index Expanded (Web of Science), and Current Contents Connect (Web of Science) up to April 24, 2020. Eligible methods studies assessed the impact of restricting systematic reviews to English-language publications on effect estimates and conclusions. Two reviewers independently screened the literature; one investigator performed the data extraction, a second investigator checked for completeness and accuracy. We synthesized the findings narratively.
Eight methods studies (10 publications) met the inclusion criteria; none addressed language restrictions in diagnostic test accuracy reviews. The included studies analyzed nine to 147 meta-analyses and/or systematic reviews. The proportions of non-English-language publications ranged from 2% to 100%. Based on five methods studies, restricting literature searches or inclusion criteria to English-language publications led to a change in statistical significance in 23/259 meta-analyses (9%). Most commonly, the statistical significance was lost, but had no impact on the conclusions of systematic reviews.
Restricting systematic reviews to English-language publications appears to have little impact on the effect estimates and conclusions of systematic reviews.

Results of meta-analyses are potentially valuable for informing environmental policy and practice decisions. However, selective sampling of primary studies through searches exclusively using widely used bibliographic platform(s) could bias estimates of effect sizes. Such search strategies are common in environmental evidence reviews, and if risk of bias can be detected, this would provide the first empirical evidence that comprehensiveness of searches needs to be improved. We compare the impact of using single and multiple bibliographic platform(s) searches vs more comprehensive searches on estimates of mean effect sizes. We used 137 published meta-analyses, based on multiple source searches, analyzing 9388 studies: 8095 sourced from commercially published articles; and 1293 from grey literature and unpublished data. Single-platform and multiple-platform searches missed studies in 100 and 80 of the meta-analyses, respectively: 52 and 46 meta-analyses provided larger-effect estimates; 32 and 28 meta-analyses provided smaller-effect estimates; eight and four meta-analyses provided opposite direction of estimates; and two each were unable to estimate effects due to missing all studies. Further, we found significant positive log-linear relationships between proportions of studies missed and the deviations of mean effect sizes, suggesting that as the number of studies missed increases, deviation of mean effect size is likely to expand. We also found significant differences in mean effect sizes between indexed and non-indexed studies for 35% of meta-analyses, indicating high risk of bias when the searches were restricted. We conclude that the restricted searches are likely to lead to unrepresentative samples of studies and biased estimates of true effects.

Studies published in languages other than English are often neglected when research teams conduct systematic reviews. Literature on how to deal with non-English studies when conducting reviews have focused on the importance of including such studies, while less attention has been paid to the practical challenges of locating and assessing relevant non-English studies. We investigated the factors which might predict the inclusion of non-English studies in systematic reviews in the social sciences, to better understand how, when and why these are included/excluded.
We appraised all Campbell Collaboration systematic reviews (n = 123) published to July 2016, categorising each by its language inclusiveness. We sought additional information from review authors via a questionnaire and received responses concerning 47 reviews. Data were obtained for 17 factors and we explored correlations with the number of non-English studies in the reviews via statistical regression models. Additionally, we asked authors to identify factors that support or hinder the inclusion of non-English studies.
Of 123 reviews, 108 did not explicitly exclude, and of these, 17 included non-English language studies. One factor correlated with the number of included non-English studies across all models: the number of countries in which the members of the review team work (B-value = 0.56; SE B = 0.24; 95% CI = 0.07-1.03; p = 0.02). This indicates that reviews which included non-English studies were more likely to be produced by international review teams. Our survey showed a dominance of researchers from English-speaking countries (52.9%) and review teams consisting only of team members from these countries (65.9%). The most frequently mentioned challenge to including non-English studies was a lack of resources (funding and time) followed by a lack of language resources (e.g. professional translators).
Our findings may indicate a connection between the limited inclusion of non-English studies and a lack of resources, which forces review teams to rely on their limited language skills rather than the support of professional translators. If unaddressed, review teams risk ignoring key data and introduce bias in otherwise high-quality reviews. However, the validity and interpretation of our findings should be further assessed if we are to tackle the challenges of dealing with non-English studies.

In experimental aesthetics the relationship between the arts and cognitive neuroscience has gained particular interest in recent years. But has cognitive neuroscience indeed something to offer when studying the arts? Here we present a theoretical frame within which the concept of complementarity as a generative or creative principle is proposed; neurocognitive processes are characterized by the duality of complementary activities like bottom-up and top-down control, or logistical functions like temporal control and content functions like perceptions in the neural machinery. On that basis a thought pattern is suggested for aesthetic appreciations and cognitive appraisals in general. This thought pattern is deeply rooted in the history of philosophy and art theory since antiquity; and complementarity also characterizes neural operations as basis for cognitive processes. We then discuss some challenges one is confronted with in experimental aesthetics; in our opinion, one serious problem is the lack of a taxonomy of functions in psychology and neuroscience which is generally accepted. This deficit makes it next to impossible to develop acceptable models which are similar to what has to be modeled. Another problem is the severe language bias in this field of research as knowledge gained in many languages over the ages remains inaccessible to most scientists. Thus, an inspection of research results or theoretical concepts is necessarily too narrow. In spite of these limitations we provide a selective summary of some results and viewpoints with a focus on visual art and its appreciation. It is described how questions of art and aesthetic appreciations using behavioral methods and in particular brain-imaging techniques are analyzed and evaluated focusing on such issues like the representation of artwork or affective experiences. Finally, we emphasize complementarity as a generative principle on a practical level when artists and scientists work directly together which can lead to new insights and broader perspectives on both sides.