Artificial Intelligence Exhibits Critical Blind Spots in Allergy Test Room Safety Design.

The human cortex encodes information in complex networks that can be anatomically dispersed and variable in their microstructure across individuals. Using simulations with neural network models, we show that contemporary statistical methods for functional brain imaging-including univariate contrast, searchlight multivariate pattern classification, and whole-brain decoding with L1 or L2 regularization-each have critical and complementary blind spots under these conditions. We then introduce the sparse-overlapping-sets (SOS) LASSO-a whole-brain multivariate approach that exploits structured sparsity to find network-distributed information-and show in simulation that it captures the advantages of other approaches while avoiding their limitations. When applied to fMRI data to find neural responses that discriminate visually presented faces from other visual stimuli, each method yields a different result, but existing approaches all support the canonical view that face perception engages localized areas in posterior occipital and temporal regions. In contrast, SOS LASSO uncovers a network spanning all four lobes of the brain. The result cannot reflect spurious selection of out-of-system areas because decoding accuracy remains exceedingly high even when canonical face and place systems are removed from the dataset. When used to discriminate visual scenes from other stimuli, the same approach reveals a localized signal consistent with other methods-illustrating that SOS LASSO can detect both widely distributed and localized representational structure. Thus, structured sparsity can provide an unbiased method for testing claims of functional localization. For faces and possibly other domains, such decoding may reveal representations more widely distributed than previously suspected.SIGNIFICANCE STATEMENT Brain systems represent information as patterns of activation over neural populations connected in networks that can be widely distributed anatomically, variable across individuals, and intermingled with other networks. We show that four widespread statistical approaches to functional brain imaging have critical blind spots in this scenario and use simulations with neural network models to illustrate why. We then introduce a new approach designed specifically to find radically distributed representations in neural networks. In simulation and in fMRI data collected in the well studied domain of face perception, the new approach discovers extensive signal missed by the other methods-suggesting that prior functional imaging work may have significantly underestimated the degree to which neurocognitive representations are distributed and variable across individuals.

Despite high unmet demand for health services across rural Australia, uptake of telehealth has been slow, piecemeal and ad hoc. We argue that widespread failure to understand telehealth as a socio-technical practice is key to understanding this slow progress. To develop this argument, we explore how technocentric approaches to telehealth have contributed to critical blind spots. First, the 'hype' associated with the technological possibilities of telehealth discourages thoughtful consideration of the unanticipated consequences when technologies are rolled out into complex social fields. Second, it contributes to critical gaps in the telehealth evidence base, and particularly a paucity of analyses focussing on the experiences of service users and patients. A third blind spot concerns the limited attention paid to the social determinants of health and digital divides in rural areas. The final blind spot we consider is an apparent reluctance to engage community stakeholders in co-designing and coproducing telehealth services. We used an iterative approach to identify studies and commentary from a range of academic fields to explain the significance of the telehealth blind spots and how they might be addressed. Insights suggest how expanding understanding of the social dimensions of telehealth could enhance its accessibility, effectiveness and responsiveness to community needs and contexts.

Artificial intelligence (AI) is accelerating discovery timelines in synthetic biology, expanding opportunities for therapeutic breakthroughs, sustainable bio-manufacturing, and rapid response to health and environmental challenges. At the same time, this acceleration shifts biosecurity risks from physical materials toward a broader socio-technical landscape involving models, datasets, and distributed automation. This literature review synthesizes evidence from 119 peer-reviewed articles published between January 2015 and August 2025, focusing on biosecurity risks, dual-use concerns, and governance responses at the AI–synthetic-biology interface. Findings indicate that AI systems consistently increase design throughput and lower expertise barriers, enabling faster medical and industrial innovation but also heightening risks of repurposing for harmful molecules or genetic sequences. Existing governance remains fragmented: biosafety regimes emphasize physical agents and laboratories, while AI governance frameworks focus on privacy and fairness—leaving critical blind spots for biological misuse scenarios. Mitigation measures identified in the literature converge on layered controls, including risk-tiered access to high-capability models, systematic red-teaming prior to release, strengthened DNA-synthesis screening (including short fragments), audit logging, secure data infrastructures, and international capacity-building. Evaluation gaps persist, and harmonized metrics—such as synthesis-screening coverage, red-team testing frequency, accredited biofoundries, and early-warning lead times—are recommended for systematic monitoring of governance effectiveness. Addressing these challenges is essential to ensure that AI-enabled synthetic biology advances responsibly, balancing its transformative potential for health and sustainability with global biosecurity.

Following a thorough examination of the State’s policies on migration control, the privatisation of migration control and the subsequent implementation by private actors, this chapter highlights the gap between the policies’ objectives and actual outcome which, inadvertently, led to the creation of blind spots that compromise and evade State control and regulation. This chapter thus provides a comprehensive and analytical view of the most critical blind spots that it believes should be addressed. It engages with both sides of the migration industry to expose the exploitation and mistreatment of vulnerable migrants and the lack of sufficient oversight and transparency. Finally, it explores the lock-in effect phenomenon and private actors’ irreversible involvement in policy-making.

SummaryBackground:Type 1 diabetes requires frequent testing and monitoring of blood glucose levels in order to determine appropriate type and dosage of insulin administration. This can lead to thousands of individual measurements over the course of a lifetime of a single individual, of which very few are retained as part of a permanent record. The third author, aged 9, and his family have maintained several years of written records since his diagnosis with Type 1 diabetes at age 20 months, and have also recently begun to obtain automated records from a continuous glucose monitor.Objectives:This paper compares regularities identified within aggregated manually-collected and automatically-collected blood glucose data visualizations by the family involved in monitoring the third author’s diabetes.Methods:7,437 handwritten entries of the third author’s blood sugar readings were obtained from a personal archive, digitized, and visualized in Tableau data visualization software. 6,420 automatically collected entries from a Dexcom G4 Platinum continuous glucose monitor were obtained and visualized in Dexcom’s Clarity data visualization report tool. The family was interviewed three times about diabetes data management and their impressions of data as presented in data visualizations. Interviews were audiorecorded or recorded with handwritten notes.Results:The aggregated visualization of manually-collected data revealed consistent habitual times of day when blood sugar measurements were obtained. The family was not fully aware that their existing life routines and the third author’s entry into formal schooling had created critical blind spots in their data that were often unmeasured. This was realized upon aggregate visualization of CGM data, but the discovery and use of these visualizations were not realized until a new healthcare provider required the family to find and use them. The lack of use of CGM aggregate visualization was reportedly because the default data displays seemed to provide already abundant information for in-the-moment decision making for diabetes management.Conclusions:Existing family routines and school schedules can shape if and when blood glucose data are obtained for T1D youth. These routines may inadvertently introduce blind spots in data, even when it is collected and recorded systematically. Although CGM data may be superior in its overall density of data collection, families do not necessarily discover nor use the full range of useful data visualization features. To support greater awareness of youth blood sugar levels, families that manually obtain youth glucose data should be advised to avoid inadvertently creating data blind spots due to existing schedules and routines. For families using CGM technology, designers and healthcare providers should consider implementing better cues and prompts that will encourage families to discover and utilize aggregate data visualization capabilities.

Disaster Risk Management (DRM) has been evolving under the pressure of new challenges brought by increasingly frequent and severe multi-hazard events. These events are more likely to impact multiple countries at once, exposing common and specific vulnerabilities of neighbouring communities. One prominent and recent example for Europe comes from the flood events in 2021, considered one of the most destructive hydrological disasters of the 21st century. In Europe, the July 2021 floods claimed over 200 lives, causing widespread disruption and economic loss exceeding 50 billion euros.The resulting shared but distinct experiences call for joint reflection from scientists and stakeholders from the impacted countries and regions – an exercise whose significance we are only beginning to understand.This study aims to cross-examine the impacts of the 2021 flood events in Romania and the Netherlands, alongside the vulnerabilities that contributed to them and the adaptation options employed to address them. Drawing from a wide range of sources (e.g., scientific papers, official reports, administrative acts, hydro-meteorological datasets, and news reports), two distinct Impact Chains were developed, one for each country. From these models, we elicited lessons regarding the best practices and blind spots in DRM.The Impact Chains revolve around the most severely affected areas in the two case studies: Alba County in the northwest of Romania and Limburg province in the southeast of the Netherlands. The chains include cascading hazards such as floods, heavy rainfall, strong winds, and landslides. To ensure their accuracy and reliability, the models were calibrated and validated through stakeholder surveys conducted in each case study area.Employing a set of Kumu metrics and other custom-designed metrics, the two Impact Chains were analysed to identify the most prominent flood impacts, vulnerabilities, and adaptation options. The comparative analysis provided key insights into the DRM approaches in Romania and the Netherlands, which were leveraged to pinpoint both strengths worth of replication and weaknesses that should be avoided. Notable best DRM practices refer to effective search and rescue operations in both countries, the simplification of flood damage compensation procedures in Romania , and the swift evacuation and accommodation of the population in Limburg. In terms of critical blind spots, both countries are yet to design (multi-)hazard management strategies that factor in pandemic conditions and that also proactively address vulnerabilities rather than merely mitigating flood impacts.These DRM lessons offer relevant answers to the crux questions that arise following major hazardous events, such as the floods of 2021: What can be done to fend off such severe impacts in the future? and What can we learn from the experience of other countries? By bringing together examples of best practices and pitfalls of DRM, this study fosters constructive dialogue grounded in shared experiences.This research opens the way to further Impact Chain-based cross-country comparisons of multi-hazards, in an effort conducive to collaboratively deciphering the interplay of multi-risk in diverse contexts and to linking it with country- or region-specific DRM policies and practices.

Chemical contaminants in the environment invariably occur as complex mixtures, yet conventional risk assessment paradigms rely on single‑chemical evaluations that cannot capture emergent synergistic or antagonistic interactions. This review critically examines how artificial intelligence (AI) integrated with New Approach Methodologies (NAMs) is fundamentally reshaping mixture toxicity assessment. We directly compare four AI architectures traditional machine learning, Graph Neural Networks (GNNs), SMILES‑based Transformers, and the highly interpretable q‑RASAR framework, critically evaluating their respective abilities to model non‑additive mixture effects, handle data scarcity, and meet regulatory interpretability standards. We address critical blind spots in emerging contaminant mixtures, including microplastics acting as vectors for antibiotics and heavy metals, and nanomaterial‑driven acceleration of antibiotic resistance gene transfer. A central theme is the indispensable transition from statistical correlation to causal biological proof: we delineate the boundary between statistical explainability (SHAP values) and mechanistic explainability (Adverse Outcome Pathways), and we propose a closed‑loop validation strategy that combines in vitro organoid assays, AOP mapping, and AI‑optimized PBPK modeling. Finally, we provide a regulatory roadmap that emphasizes globally standardized validation protocols (e.g., the OECD (Q)SAR Assessment Framework), large‑scale chemistry‑biology interaction databases for zero‑shot mixture predictions, Bayesian uncertainty quantification, and legal frameworks for liability when using black‑box models.

Purpose Crystal-clear visibility is the lifeblood of competitive advantage, yet “visibility hotspots” act as choke points, strangling information flow and crippling supply chains. This study delves into these critical blind spots, revealing their manifestation and threats to performance, paving the way for a smoother, more competitive supply chain. Design/methodology/approach Through multiple real-world case studies and semi-structured interviews with supply chain managers, this study explores the factors behind these visibility hotspots and their crippling consequences. A fuzzy visibility hotspots assessment method is introduced to facilitate the exploration. Findings The study confirms “visibility hotspots” as significant roadblocks in industrial management, with diverse causes linked to both supply chain processes and visibility influencing factors. Importantly, supply chain managers see the innovative “fuzzy assessment method” as a key tool to identify and tackle these blind spots, paving the way for enhanced competitive advantage. Originality/value This study introduces a novel methodology for identifying supply chain visibility hotspots, empowering practitioners to gain unprecedented clarity across their entire chain. This unlocks critical insights, leading to enhanced visibility, more efficient resource management and a stronger competitive edge.

The dominance of conflict theory in contemporary sociology has contributed to critical blind spots, including the discipline’s limited attention to social progress and the conditions that foster it. This paper traces the rise and fall of functionalism, examines current blind spots in sociology such as rising living standards and declines in racial and gender inequality, and identifies the theoretical and ideological factors that contribute to these omissions. The discipline can be better equipped to explain social progress by using a broader array of theoretical tools and embracing greater ideological and moral pluralism. Doing so would make sociology more relevant to public conversations about important social issues that will resonate with people from across the political spectrum and increase public trust in sociological teaching and research.

Text-to-image (T2I) generation, a core component of generative artificial intelligence(AI), is increasingly important for creative industries and human–computer interaction. Despite impressive progress in realism and diversity, diffusion models still exhibit critical security blind spots particularly in the Transformer key-value mapping mechanism that underpins cross-modal alignment. Existing backdoor attacks often rely on large-scale data poisoning or extensive fine-tuning, leading to low efficiency and limited stealth. To address these challenges, we propose two efficient backdoor attack methods AttnBackdoor and SemBackdoor grounded in the Transformer’s key-value storage principle. AttnBackdoor injects precise mappings between trigger prompts and target instances by fine-tuning the key-value projection matrices in U-Net cross-attention layers (≈5% of parameters). SemBackdoor establishes semantic-level mappings by editing the text encoder’s MLP projection matrix (≈0.3% of parameters). Both approaches achieve high attack success rates (>90%), with SemBackdoor reaching 98.6% and AttnBackdoor 97.2%. They also reduce parameter updates and training time by 1–2 orders of magnitude compared to prior work while preserving benign generation quality. Our findings reveal dual vulnerabilities at visual and semantic levels and provide a foundation for developing next generation defenses for secure generative AI.

Assessing perspectives on artificial intelligence applications to gastroenterology

Reviewed by: Science in Modern Poetry: New Directions ed. by John Holmes Katherine Ebury Science in Modern Poetry: New Directions. John Holmes, ed. Liverpool: Liverpool University Press, 2012. Pp. vii + 237. $99.95 (cloth). Science in Modern Poetry, a collection of twelve essays by leading critics on modern poetry and on literature and science, amply addresses two specific critical blind spots. In the volume’s editorial introduction, John Holmes points out that, so far, studies of literature and science have concentrated far more on earlier literary periods and on prose texts, leaving modern and contemporary poetry doubly neglected. Furthermore, this collection challenges the relevance of the “two cultures” debate for contemporary literature; Snow’s idea, though discredited, might still keep talented students (and even researchers) from embarking on interdisciplinary projects. As Holmes puts it, “If C. P. Snow’s famous diagnosis of a rift between the ‘two cultures’ of science and letters has rarely been applicable to modern poets, today his model looks less apt than ever” (3). It can only be helpful to point this out in a scholarly way. [End Page 159] The aim of the collection is “to give to scholars and students approaching the topic of science in modern poetry from either side—from an interest either in literature and science or in modern poetry per se” (5). It is also possible to hope that the volume could also address another audience: scientists interested in poetry. After all, the index allows one to search “by poet” or “by scientist/science,” and the introduction gives examples of scientists who have cited poetry in their work (including Edward Wilson, Richard Dawkins and James Lovelock, and Nicholas Battey). To that end, the volume would be more interesting if it included a contemporary scientist’s perspective on twentieth-century poetry, if only as a counter to C. P. Snow. To Holmes’s credit, though, he did secure the perspectives of two contemporary poets, John Barnie and Robert Crawford, whose contributions enrich the collection. Holmes, along with the volume’s contributors, remedies the critical neglect of poetry and the twentieth century within the study of literature and science. He accomplishes this, not by offering a theoretical primer for interdisciplinary studies of modern and contemporary poetry, but by providing a more engaging record of current research projects in the field. The book “embod[ies] certain trends in research on science and modern poetry” and points out new directions for future study (10). This means that its arrangement is not tied to chronology. Instead, it is divided into three specific sections, “Science and Contemporary Poetry,” “Science in Modernist Poetry,” and “Darwinian Dialogues.” This arrangement seems to give considerable weight to Darwin, whose section is by far the strongest and most coherent. In fact, Holmes frames the collection by arguing that we now see a “biological turn” in scholarship on modern and contemporary literature (15), making this final section both the heart of the volume and the last word on the subject. While it would be impossible to dispute that much exciting work is going on in this area of Darwinian interpretations of literature and science, other biology-focused research is necessarily unrepresented (for example, the turn towards neuroscience in modernist studies, as evidenced by several recent conferences). Despite the individual strength of each of the chapters and the interesting diversity of sciences invoked, it would be interesting to envisage a volume that either dealt specifically with Darwin in modern poetry or that did not emphasize one specific trend in research over other trends. Nonetheless, each section is rich and complex in unique ways. The first, “Science and Contemporary Poetry,” contains some fascinating essays: Helen Small examines the dual status (poet/scientist) shared by Miroslav Holub and Roald Hoffmann; Peter Middleton analyzes language writing and molecular biology; John Barnie looks at Edward O. Wilson and A. R. Ammons; and Robert Crawford discusses poetry and science in the contemporary university. Small’s and Crawford’s contributions are particularly interesting because they deal with direct interchange between poetry and science; both Holub and Hoffmann are poet-scientists (or scientist-poets), and Crawford’s essay offers an account of the poet Michael Donaghy meeting Kevin Warwick, the...

This review reconceptualises internalised misogyny as a clinically relevant mechanism of psychological distress in women, rather than a peripheral sociopolitical phenomenon. Synthesising feminist psychological theory with clinical research, it investigates how internalised sexist beliefs manifest as shame, perfectionism, emotional suppression, disordered eating, and trauma-related symptoms. The paper critiques the Internalised Misogyny Scale for lacking cultural and intersectional sensitivity, rendering them inadequate in diverse contexts like India. Diagnostic frameworks such as the DSM and ICD similarly omit sexism as etiological, leading to fragmented and depoliticised clinical interpretations. Drawing on Indian feminist scholarship, the paper explores how caste, colourism, respectability politics, and religious norms reinforce gendered self-policing. It exposes critical blind spots in mainstream and even feminist clinical models, and argues for feminist-informed therapies, such as narrative and relational-cultural therapy, that validate gendered suffering and challenge oppressive internalised scripts. By framing internalised misogyny as a somatised, culturally entangled, and under-recognised driver of women’s psychopathology, this review advocates for decolonial, trauma-informed clinical practices and culturally grounded future research.

A systematic review of traffic safety data collection methods and challenges: From crash databases to AI-augmented sensors.

This chapter thinks through international law and posthuman theory by way of practice of ‘posthumanist commoning’. It explores the posthumanist and the commoning dimensions of particular legal and political collective actions at hand. It does so by telling the story of the ‘insurgent lake’ of Rome – the ‘lago bullicante’. Bullicante is an archaic Italian term that signifies both ‘to boil’ (bollire) and ‘to get agitated’ (agitarsi). The ‘lake that boils and gets agitated’ refers to the artificial/natural lake that was accidentally created in 1992, when an underground parking lot was illegally constructed and inadvertently hit an aquifer, thereby flooding the construction site and nearby area, creating a one-hectare large lake in the heart of the city. With the lake, an insurgent political subjectivity emerged to resist and care for its preservation. Both the subjectivity and the struggle are articulated and practiced in non-liberal, non-individualistic, and in-human (or more and less than ‘human’) terms, thereby giving rise to a distinctive mode of ‘becoming common’. Drawing on the lago bullicante, I argue that this mode of ‘posthumanist commoning’ enacts particular practices of ecological resistance, refusal and reparation. The transversal alliances forged within networks of transnational resisting collectives help exploring how posthuman theory can inform international law. It does so by availing methods of reconfiguring the categories of the human, the land and its living ecology, while also revealing critical blind spots and methodological/conceptual limitations of both posthuman theory and international law.