Explore the vast range of titles available.

Accurate emotion recognition is a foundational component of social cognition, yet human biases can compromise its reliability. The emergent capabilities of multimodal large language models (MLLMs) offer a potential avenue for objective analysis, but their performance has been tested mainly with ethnically homogenous stimuli. This study provides a systematic cross-ethnic evaluation of leading MLLMs on an emotion recognition task to assess their accuracy and consistency across diverse groups. We evaluated three leading MLLMs: ChatGPT-4, ChatGPT-4o, and Claude 3 Opus. Performance was tested twice using three “Reading the Mind in the Eyes Test” (RMET) versions featuring White, Black, and Korean faces. We analyzed accuracy against chance (25%) and compared scores to established human normative data for each ethnic version. ChatGPT-4o achieved performance significantly above chance levels across all tests ( p < .001), with large effect sizes indicating robust performance (Cohen’s h = 1.253–1.619; RD = 0.583–0.694). The model obtained a mean accuracy of 83.3% (30/36) on the White RMET, 94.4% (34/36) on the Black RMET, and 86.1% (31/36) on the Korean RMET, placing it in the 85th, 94th, and 90th percentiles of human norms, respectively. This high accuracy remained consistent across ethnic stimuli. In contrast, ChatGPT-4 performed near the human average, while Claude 3 Opus performed near chance level. These preliminary findings highlight the rapid evolution of MLLMs, highlighting a significant performance leap between consecutive versions. This study suggests that ChatGPT-4o demonstrated performance scores exceeding average human accuracy on this specific task in recognizing complex emotions from static images of the eye region, with its performance remaining consistent across different ethnic groups. While these results are notable, the pronounced performance gaps between models and the inherent limitations of the RMET task underscore the need for continuous validation and careful, ethical consideration to fully understand the capabilities and boundaries of this technology.

Background Despite significant advances in AI-driven medical diagnostics, the integration of large language models (LLMs) into psychiatric practice presents unique challenges. While LLMs demonstrate high accuracy in controlled settings, their performance in collaborative clinical environments remains unclear. This study examined whether LLMs exhibit conformity behavior under social pressure across different diagnostic certainty levels, with a particular focus on psychiatric assessment. Methods Using an adapted Asch paradigm, we conducted a controlled trial examining GPT-4o’s performance across three domains representing increasing levels of diagnostic uncertainty: circle similarity judgments (high certainty), brain tumor identification (intermediate certainty), and psychiatric assessment using children’s drawings (high uncertainty). The study employed a 3 × 3 factorial design with three pressure conditions: no pressure, full pressure (five consecutive incorrect peer responses), and partial pressure (mixed correct and incorrect peer responses). We conducted 10 trials per condition combination (90 total observations), using standardized prompts and multiple-choice responses. The binomial test and chi-square analyses assessed performance differences across conditions. Results Under no pressure, GPT-4o achieved 100% accuracy across all domains. Under full pressure, accuracy declined systematically with increasing diagnostic uncertainty: 50% in circle recognition, 40% in tumor identification, and 0% in psychiatric assessment. Partial pressure showed a similar pattern, with maintained accuracy in basic tasks (80% in circle recognition, 100% in tumor identification) but complete failure in psychiatric assessment (0%). All differences between no pressure and pressure conditions were statistically significant ( P  <.05), with the most severe effects observed in psychiatric assessment (χ²₁=16.20, P  <.001). Conclusions This study reveals that LLMs exhibit conformity patterns that intensify with diagnostic uncertainty, culminating in complete performance failure in psychiatric assessment under social pressure. These findings suggest that successful implementation of AI in psychiatry requires careful consideration of social dynamics and the inherent uncertainty in psychiatric diagnosis. Future research should validate these findings across different AI systems and diagnostic tools while developing strategies to maintain AI independence in clinical settings. Trial registration Not applicable.

JEFFREY COLIN RICHARDS CLINICAL PSYCHOLOGIST 16-6-1949 - 5-4-2005

Psychosocial interventions have the potential to enhance relapse prevention in bipolar disorder. To evaluate a manualised group-based intervention for people with bipolar disorder in a naturalistic setting. Eighty-four participants were randomised to receive the group-based intervention (a 12-week programme plus three booster sessions) or treatment as usual, and followed up with monthly telephone interviews (for 9 months post-intervention) and face-to-face interviews (at baseline, 3 months and 12 months). Participants who received the group-based intervention were significantly less likely to have a relapse of any type and spent less time unwell. There was a reduced rate of relapse in the treatment group for pooled relapses of any type (hazard ratio 0.43, 95% CI 0.20-0.95; t(343) = -2.09, P = 0.04). This study suggests that the group-based intervention reduces relapse risk in bipolar disorder.

The stepped care model for psychological distress has been promoted in recent years, leading to the enhancing roles of primary care professionals and alternative sources of help. However, most of the research findings come from Western countries. This study investigates help-seeking behaviours of Chinese patients among different types of professional and alternative sources for psychological distress in Hong Kong. A questionnaire survey was conducted with 1626 adult primary care attenders from 13 private and 6 public clinics, 650 (40.0%) reported that they had ever experienced psychological distress. Their help-seeking behaviours, demographic background and current distress level (measured by GHQ-12) were analysed. Among the respondents with experience of psychological distress, 48.2% had sought help from professional and/or alternative sources for their distress [10.2% from professionals only, 12.6% from alternative sources only, and 25.4% from both]. Those who had sought help from professionals only were more likely to be less educated and with lower income. In contrast, those using alternative sources only were more likely to be younger, better educated, and have higher income. Allowing multiple responses, psychiatrists (22.3%) was reported to be the most popular professional source, followed by primary care physicians (17.5%), clinical psychologists (12.8%) and social workers/counsellors (12.0%). Family members/friends (28.6%) was the top alternative source, followed by exercise/sports (21.8%), religious/spiritual support (16.9%) and self-help websites/books/pamphlets (8.9%). While psychiatrists remain the most popular professional source of help to the Chinese patients in Hong Kong, primary care professionals and alternative sources also play significant roles. Distressed patients who are younger, better educated and have higher income are more likely to use alternative sources only. The outcomes need further research.

Studies of Caenorhabditis elegans vulval development provide a paradigm for pattern formation during animal development. The fates of the six vulval precursor cells are specified by the combined action of an inductive signal that activates the EGF receptor mitogen-activated PK signaling pathway (specifying a primary fate) and a lateral signal mediated by LIN-12/Notch (specifying a secondary fate). Here we use methods devised for the engineering of complex reactive systems to model a biological system. We have chosen the visual formalism of statecharts and use it to formalize Sternberg and Horvitz's 1989 model [Sternberg, P. W. & Horvitz, H. R. (1989) Cell 58, 679-693], which forms the basis for our current understanding of the interaction between these two signaling pathways. The construction and execution of our model suggest that different levels of the inductive signal induce a temporally graded response of the EGF receptor mitogen-activated PK pathway and make explicit the importance of this temporal response. Our model also suggests the existence of an additional mechanism operating during lateral specification that prohibits neighboring vulval precursor cells from assuming the primary fate.

Chronic Myeloid Leukemia (CML) represents a paradigm for the wider cancer field. Despite the fact that tyrosine kinase inhibitors have established targeted molecular therapy in CML, patients often face the risk of developing drug resistance, caused by mutations and/or activation of alternative cellular pathways. To optimize drug development, one needs to systematically test all possible combinations of drug targets within the genetic network that regulates the disease. The BioModelAnalyzer (BMA) is a user-friendly computational tool that allows us to do exactly that. We used BMA to build a CML network-model composed of 54 nodes linked by 104 interactions that encapsulates experimental data collected from 160 publications. While previous studies were limited by their focus on a single pathway or cellular process, our executable model allowed us to probe dynamic interactions between multiple pathways and cellular outcomes, suggest new combinatorial therapeutic targets and highlight previously unexplored sensitivities to Interleukin-3.

We present a translation from linear temporal logic with past to deterministic Rabin automata. The translation is direct in the sense that it does not rely on intermediate non-deterministic automata, and asymptotically optimal, resulting in Rabin automata of doubly exponential size. It is based on two main notions. One is that it is possible to encode the history contained in the prefix of a word, as relevant for the formula under consideration, by performing simple rewrites of the formula itself. As a consequence, a formula involving past operators can (through such rewrites, which involve alternating between weak and strong versions of past operators in the formula's syntax tree) be correctly evaluated at an arbitrary point in the future without requiring backtracking through the word. The other is that this allows us to generalize to linear temporal logic with past the result that the language of a pure-future formula can be decomposed into a Boolean combination of simpler languages, for which deterministic automata with simple acceptance conditions are easily constructed.

In an age where the volume of data regarding biological systems exceeds our ability to analyse it, many researchers are looking towards systems biology and computational modelling to help unravel the complexities of gene and protein regulatory networks. In order to make such techniques more accessible to mainstream researchers, tools such as the BioModelAnalyzer (BMA) have been developed to provide a user-friendly graphical interface for discrete modelling of biological systems. Here we use the BMA to build a library of target functions of known molecular interactions, translated from ordinary differential equations (ODEs). We then show that these BMA target functions can be used to reconstruct complex networks, which can correctly predict many known genetic perturbations. This new library supports the accessibility ethos behind the creation of BMA, providing a toolbox for the construction of complex cell signalling models without the need for extensive experience in computer programming or mathematical modelling, and allows for construction and simulation of complex biological systems with only small amounts of quantitative data.