Discover Research
Explore trending papers converted into audio discussions.
Mammary Fibroblasts Secrete Damage Associated Molecular Patterns through Extracellular Vesicles in Response to Ionizing Radiation
Ionizing radiation (IR) is an integral component of cancer therapy. Cellular exposure to IR typically leads to major biological consequences including cell death and senescence. Furthermore, tissue injury in known to involve the release of damage-associated molecular patterns (DAMPs) into the extracellular space, which trigger inflammation and wound healing. However, DAMP release in the context of radiation injury remains to be fully characterized. Evidence suggests that extracellular vesicle (EV) secretion and associated cargo components are part of the cellular response to IR, but the mechanisms integrating cellular damage and EV secretion post-IR are largely unexplored. In this study, we show that acute IR-induced damage in mammary fibroblasts results in a senescence-like phenotype and substantially increased EV secretion. Quantitative proteomic analysis revealed that IR-induced EVs are enriched with extracellular and intracellular DAMPs, along with other pro-inflammatory mediators. We show that knockdown of the GTPase Rab27a abrogates IR-induced EV secretion and inhibits the enrichment of key DAMPs in EVs. By examining the integration of cellular damage and senescence with the release of inflammatory signals, this study elucidates a potentially critical role for EV-associated proteins in the radiation response.
Nanometer-precision tracking of adipocyte dynamics via single lipid droplet whispering‑gallery optical resonances
Biophotonics—and more recently, biointegrated photonics—offer transformative tools for probing cellular processes with unprecedented precision. Among these, whispering gallery mode (WGM) resonators—optical microcavities formed in spherical structures—have emerged as powerful biosensors and intracellular barcodes. Lipid droplets (LDs), with their high refractive index and intrinsic spherical geometry, are ideal candidates for supporting intracellular lasing. Although lasing in LDs has been previously demonstrated, it has not yet been harnessed to study live cell biology. Here, we report the first use of WGM resonances in LDs of live primary adipocytes, employing a continuous-wave (CW) laser at powers below the biological damage threshold. By measuring these resonances, we achieved nanometer-scale precision in size estimation, enabling real-time observation of rapid LD dynamics and deformations on the minute scale—far beyond the spatio-temporal resolution of conventional microscopy. We systematically characterized this photonic sensing approach, demonstrating its ability to resolve adipocyte heterogeneity, monitor lipolytic responses to forskolin and isoproterenol, and detect early signs of cell viability loss—well before conventional assays. This proof-of-concept establishes intracellular LD WGM resonances as a robust platform for investigating live single-cell metabolism. The technique enables rapid, cost-effective assessment of adipocyte function, reveals cell-to-cell variability obscured by bulk assays, and lays the foundation for high-throughput analysis of metabolism- and obesity-related diseases at both cellular and tissue levels.
Human-Centred Evaluation of Text-to-Image Generation Models for Self-expression of Mental Distress: A Dataset Based on GPT-4o
Effective communication is central to achieving positive healthcare outcomes in mental health contexts, yet international students often face linguistic and cultural barriers that hinder their communication of mental distress. In this study, we evaluate the effectiveness of AI-generated images in supporting self-expression of mental distress. To achieve this, twenty Chinese international students studying at UK universities were invited to describe their personal experiences of mental distress. These descriptions were elaborated using GPT-4o with four persona-based prompt templates rooted in contemporary counselling practice to generate corresponding images. Participants then evaluated the helpfulness of generated images in facilitating the expression of their feelings based on their original descriptions. The resulting dataset comprises 100 textual descriptions of mental distress, 400 generated images, and corresponding human evaluation scores. Findings indicate that prompt design substantially affects perceived helpfulness, with the illustrator persona achieving the highest ratings. This work introduces the first publicly available text-to-image evaluation dataset with human judgment scores in the mental health domain, offering valuable resources for image evaluation, reinforcement learning with human feedback, and multi-modal research on mental health communication. Keywords: Text-to-Image Generation, Human Generated & Evaluated Dataset, Mental Health, Self-Expression
Human-Centred Evaluation of Text-to-Image Generation Models for Self-expression of Mental Distress: A Dataset Based on GPT-4o
Effective communication is central to achieving positive healthcare outcomes in mental health contexts, yet international students often face linguistic and cultural barriers that hinder their communication of mental distress. In this study, we evaluate the effectiveness of AI-generated images in supporting self-expression of mental distress. To achieve this, twenty Chinese international students studying at UK universities were invited to describe their personal experiences of mental distress. These descriptions were elaborated using GPT-4o with four persona-based prompt templates rooted in contemporary counselling practice to generate corresponding images. Participants then evaluated the helpfulness of generated images in facilitating the expression of their feelings based on their original descriptions. The resulting dataset comprises 100 textual descriptions of mental distress, 400 generated images, and corresponding human evaluation scores. Findings indicate that prompt design substantially affects perceived helpfulness, with the illustrator persona achieving the highest ratings. This work introduces the first publicly available text-to-image evaluation dataset with human judgment scores in the mental health domain, offering valuable resources for image evaluation, reinforcement learning with human feedback, and multi-modal research on mental health communication.
Decentralized Social Media and Artificial Intelligence in Digital Public Health Monitoring
Digital public health monitoring has long relied on data from major social media platforms. Twitter was once an indispensable resource for tracking disease outbreaks and public sentiment in real time. Researchers used Twitter to monitor everything from influenza spread to vaccine hesitancy, demonstrating that social media data can serve as an early-warning system for emerging health threats. However, recent shifts in the social media landscape have challenged this data-driven paradigm. Platform policy changes, exemplified by Twitter's withdrawal of free data access, now restrict the very data that fueled a decade of digital public health research. At the same time, advances in artificial intelligence, particularly large language models (LLMs), have dramatically expanded our capacity to analyze large-scale textual data across languages and contexts. This presents a paradox: we possess powerful new AI tools to extract insights from social media, but face dwindling access to the data. In this viewpoint, we examine how digital public health monitoring is navigating these countervailing trends. We discuss the rise 1 of decentralized social networks like Mastodon and Bluesky as alternative data sources, weighing their openness and ethical alignment with research against their smaller scale and potential biases. Ultimately, we argue that digital public health surveillance must adapt by embracing new platforms and methodologies, focusing on common diseases and broad signals that remain detectable, while advocating for policies that preserve researchers' access to public data in privacy-respective ways.