159370 research outputs found
Sort by
Large language model-based multiagent collaboration for abstract screening toward automated systematic reviews
Objective
Systematic reviews (SRs) are essential for evidence-based practice but remain labor-intensive, especially during abstract screening. This study evaluates whether multiple large language model (multi-LLM) collaboration can improve the efficiency and reduce costs for abstract screening.
Methods
Abstract screening was framed as a question-answering (QA) task using cost-effective LLMs. Three multi-LLM collaboration strategies were evaluated, including majority voting by averaging opinions of peers, multi-agent debate (MAD) for answer refinement, and LLM-based adjudication against answers of individual QA baselines. These strategies were evaluated on 28 SRs of the CLEF eHealth 2019 Technology-Assisted Review benchmark using standard performance metrics such as Mean Average Precision (MAP) and Work Saved over Sampling at 95% recall (WSS@95%).
Results
Multi-LLM collaboration significantly outperformed QA baselines. Majority voting was overall the best strategy, achieving the highest MAP 0.462 and 0.341 on subsets of SRs about clinical intervention and diagnostic technology assessment, respectively, with WSS@95% 0.606 and 0.680, enabling in theory up to 68% workload reduction at 95% recall of all relevant studies. MAD improved weaker models most. Our own adjudicator-as-a-ranker method was the second strongest approach, surpassing adjudicator-as-a-judge, but at a significantly higher cost than majority voting and debating.
Conclusion
Multi-LLM collaboration substantially improves abstract screening efficiency, and the success lies in model diversity. Making the best use of diversity, majority voting stands out in terms of both excellent performance and low cost compared to adjudication. Despite context-dependent gains and diminishing model diversity, MAD is still a cost-effective strategy and a potential direction of further research
Accurate quantification of steps from multiple smartphone positions
A key challenge in smartphone-based assessment of motor capacity is that patients may wear their smartphone in varying positions, while state-of-the-art algorithms have not been designed and validated for multiple device locations. This paper proposes a solution to estimating foot-to-ground initial contacts (ICs) during gait using inertial measurement unit (IMU) sensor data collected from a smartphone agnostic of its location in a cloth front or back pocket. FAIR-Q, an algorithm originally validated for data collected from the lower trunk was further tuned for this intended use, and tested on IMU data collected in cloth pocket positions from both healthy adults (n = 83, age range: 20–83 y.o.) and people with Multiple Sclerosis (n = 50, age range: 22–61 y.o., EDSS score: 0–6) during a 30s walk test. The performance of FAIR-Q was compared against a gold standard multi-sensor device in terms of sensitivity and measurement error in identifying ICs and measuring step and stride durations. Excellent performance was achieved for both groups in all tested conditions (test-level relative errors for duration measures < 1%) and using data from a large variety of smartphone devices, supporting the method’s suitability for high-frequency smartphone-based assessment of gait capacity in neurological diseases
Scottish nearshore monothalamid Foraminifera (Rhizaria): Description of five new species and one new genus
Henderson (2023) gave informal descriptions of several soft-walled, monothalamid foraminifera from intertidal zones in the Lorne area of northwest Scotland based on morphology. In the present study, we use a combination of morphological and molecular data to formally establish one new genus and five new monothalamid species from the same area. Lorneia sphaerica gen. & sp. nov. (monothalamid Clade D) has a spherical, coarsely agglutinated test containing magnetic particles and minute aperture-like openings distributed around the test. Lorneia ovalis gen. & sp. nov. (Clade D) has similar characteristics, but the test is oval, and there is a terminal aperture situated at each end. Psammophaga owensi sp. nov. (Clade E) has an oval, finely agglutinated test with a simple terminal aperture and intracellular magnetic particles. In Hilla brevis sp. nov. (Clade Y), the test is broadly oval and finely agglutinated with a reflective sheen and a large terminal aperture with a pronounced collar. Flaviatella zaninettiae sp. nov. (Clade Y) has an elongate, finely agglutinated test with a reflective sheen, a tubular terminal apertural structure, and distinctive yellow cytoplasm. Two species, Flexammina islandica Voltski and Pawlowski, 2015 and Ovammina opaca Dahlgren, 1962, are reported for the first time in Scottish coastal waters. This study underlines the importance and diversity of monothalamid foraminifera in coastal settings
Beyond barriers:a qualitative evidence synthesis protocol of South Asian involvement in mental health research in high-income countries
Abstract Introduction There is a recognised need to provide equity in healthcare through inclusion of ethnic minorities in research. South Asians, the largest ethnic minority group in the UK, experience high levels of mental illness, often against the backdrop of socioeconomic deprivation and discrimination. The research community recognises that South Asian communities are often excluded from health research. Efforts have been made to understand the barriers and facilitators to their participation. However, the participation of individuals at the intersection of mental ill health and South Asian ethnicity remains understudied. This is a protocol for the synthesis of qualitative evidence from mental health research studies about participation, motivation and barriers to recruitment of South Asians in high-income countries. Methods and analysis We will search 10 databases for qualitative evidence on participation of South Asian individuals in mental health research studies in high-income countries: MEDLINE, EMBASE, PsycINFO, Health Management Information Consortium, Social Policy and Practice, Cochrane Database of Systematic Reviews, Cochrane Central Register of Controlled Trials, Social Sciences Citation Index, Science Citation Index Expanded and ProQuest Dissertations and Theses Citation Index. Search terms for the following five concepts will be used: (1) South Asia, (2) ethnic minority, (3) mental health conditions, (4) barriers and facilitators to participation and (5) high-income countries. No date restriction will be applied to the search. Searches will be limited to studies in the English language. Study selection and data extraction will be performed by two researchers independently, using Covidence. Demographic data, thematic outputs and salient discussion points will be extracted from final full-text inclusions and entered into NVivo for coding. Meta-ethnographic approaches using first and second-order constructs from included studies will be used to form third-order constructs. This synthesis will generate new knowledge regarding the intersectionality of mental health and South Asian ethnicity. Ethics and dissemination Ethical approval is not required as this study is a synthesis of published data. This review will provide new knowledge regarding the requirements for researchers and practitioners to advance the involvement of South Asian populations in mental health research. This will undoubtedly enhance equity in the long term, reduce the burden of mental disorders and enable the provision of more effective mental health care for South Asian communities. PROSPERO registration number CRD42025626382
Blurring migrant solidarity: Navigating the troubled regime of ‘helping others’ from the urban ground of Palermo
This paper explores migrant solidarity in the aftermath of the 2015 so-called ‘migration crisis’ from the ground of the Southern Italian ‘solidarity city’ of Palermo, building on ethnographic material collected across 2022/2023. It argues that migrant solidarity both resists and reproduces racialised borders, as it is never entirely and clearly detached from the border regime, but is deeply entangled within it. Although this argument is not new per se, this paper adds empirical depth to it following two strands. First, presenting solidarity in Palermo as embedded within the city's political economy, the paper reflects on how, within a persisting unequal and racialised access to rights, solidarity can allow for multiple extractions to be operated against migrants, while simultaneously providing a ‘safer space’ for existence. Second, drawing on the experiences of migrants who have—at least from a surface level outside perspective—comparatively better positions within the city, the paper defies their portrayals as mere recipients of help, and explores the complex ways in which they navigate the restricted opportunities opened up by solidarity alongside ubiquitous borders. The paper thus shows (1) how in Palermo solidarity has produced several outcomes—moral recognition abroad, local economies, opportunities for a vast array of people involved with migration—in the overall permanence of local shortages and everyday structural racism; (2) how the local solidarity network fills the gaps of a local retreating social care and of the state's repressive politics; (3) how this network is strategically navigated by racialised migrants to carve out opportunities, social ties and space for political development
Selective empathy and the genocide in Gaza:the silence of health and academic associations
Background: Genocide is one of the most extreme forms of global health crisis. The ongoing mass atrocities in Gaza have resulted in a sharp decline in life expectancy, the systematic destruction of healthcare infrastructure and the highest number of medical personnel killed in any conflict ever recorded. While multiple human rights organisations have recognised these conditions as genocide, major health and academic associations have responded inconsistently. Main body: In response to widespread institutional inaction, an open letter was launched to urge academic and health associations to recognise the genocide in Gaza. This initiative gathered over 15,000 signatures from academics, health professionals and members of the public. It also prompted new statements of recognition from several major public health bodies, including the European Public Health Alliance, the European Public Health Association, and the World Federation of Public Health Associations. Endorsements from national societies and local healthcare authorities in Italy also demonstrated the effectiveness of public health advocacy in confronting moral silence. Nevertheless, many medical and public health societies, particularly in the United States, have either remained silent or issued neutral statements. There were stark differences in the language and moral framing of responses to Gaza compared to those to the Russian invasion of Ukraine, reflecting a phenomenon of “selective empathy”, whereby institutional solidarity varies according to national, ethnic, or geopolitical alignment. Conclusions: The silence or equivocation of many health and academic institutions in the face of genocide undermines public trust and the ethical foundations of global health. Addressing this requires global health organisations to move beyond neutrality and engage in principled advocacy to reaffirm their moral and scientific duty to defend human life and health without discrimination
Halting predicted vertebrate declines requires tackling multiple drivers of biodiversity loss
Conservation policies aiming to halt biodiversity loss often focus on globally prevalent threats like habitat loss and exploitation, yet direct and interactive effects of multiple threats remain poorly quantified. Here, we go beyond prior meta-analyses or species-level studies by providing a global, population-level empirical analysis of threat interactions by examining 3129 vertebrate population time series worldwide with documented exposure to single and multiple threats. Populations affected solely by habitat loss or exploitation do not exhibit the steepest declines; instead, disease, invasive species, pollution, and climate change are associated with faster declines. Interactive threats contribute more to population declines than temporal or spatial variation. Counterfactual analyses reveal that mitigating multiple threats is essential to achieving nonnegative vertebrate population trends and halting biodiversity loss
“…it is not the sickness itself that kills. It is the emotional trauma”: a qualitative study of the lived experience and systemic barriers of women with breast cancer in Nigeria
Background
Breast cancer is the leading cancer among Nigerian women and is commonly diagnosed at an advanced stage; yet the everyday realities that influence patient outcomes and well-being remain underexplored. Understanding the lived experiences of Nigerian breast cancer patients is critical for developing culturally appropriate, supportive, and palliative care services.
Aim
To explore the lived experiences, needs and coping strategies of women living with and beyond breast cancer in Nigeria.
Design
Qualitative study using a constructivist–interpretivist paradigm. Semi-structured, in-depth interviews were conducted online via Microsoft Teams, audio and/or video-recorded, and transcribed verbatim. Data were analysed inductively using the Framework Method, with double coding to ensure rigour.
Setting/participants
Twenty-eight women (aged 29–63 years; mean 42 years) receiving care at two oncology centres were purposively sampled between June 2024 and March 2025. Most had stage II or III disease and had lived with cancer for at least 6–12 months.
Results
Four inter-linked themes described the women’s experiences: (1) “For several months, I saw myself like a ghost amongst people”—emotional, physical, economic and social upheaval following diagnosis; (2) “But I have to talk to myself, I need to encourage myself to keep going”—personal, spiritual and peer-based coping resources; (3) “I just know that there should be a lot more awareness” of the care continuum—structural, financial and informational barriers, including catastrophic out-of-pocket costs and fragmented care pathways; (4) “It is the emotional trauma”—reframing illness, personal growth and an expressed desire to support other patients. Across the themes, women stressed unmet psychosocial needs, reliance on faith communities, and the paucity of formal peer-support or counselling services.
Conclusions
Participants reported navigating a complex interplay of financial toxicity, systemic delays, and profound psychosocial distress, and many described drawing on spiritual practices, self-encouragement, and peer connections to cope; several also expressed a desire to support other women by sharing advice or lived experience while going through the care pathway. Strengthening palliative and supportive care in Nigeria should prioritise: (1) financial-protection mechanisms to reduce treatment abandonment; (2) streamlined referral/navigation systems; and (3) integrated psychosocial and peer-support interventions that acknowledge spiritual coping. Further research should investigate survivor-led support models and assess the effectiveness of culturally tailored communication training for oncology teams
Prognostic and immunological significance of tryptophan metabolic enzymes across endometrial cancer molecular subtypes
Background
The kynurenine pathway is a key immunosuppressive mechanism implicated in resistance to immune checkpoint inhibitors (ICIs). This study investigated expression of tryptophan-metabolising enzymes (IDO1, TDO2, IL4I1) and their relationship with the immune microenvironment across molecular subtypes of endometrial cancer (EC).
Methods
A cohort of 570 ECs was classified as mismatch repair-deficient (MMRd), p53-mutant (p53mut), or no specific molecular profile (NSMP). Expression of IDO1, TDO2, IL4I1, PD-L1, kynurenine, and immune markers (CD8, FOXP3, CD68, CD163) was assessed by immunohistochemistry and quantified in tumour and stromal compartments using QuPath. Associations with disease-specific survival (DSS) were analysed using correlation testing, Kaplan–Meier, and Cox regression.
Results
IDO1, TDO2, and IL4I1 correlated strongly with immune infiltrate density. High IDO1 tumour expression was linked to improved DSS in NSMP and p53mut tumours (p < 0.05), consistent with an “inflamed” phenotype. In contrast, high TDO2 stromal expression predicted reduced DSS in NSMP patients (p < 0.05). Elevated IL4I1 tumour expression was associated with improved DSS in MMRd tumours (p < 0.05). A high CD163:CD8 ratio independently predicted worse DSS in p53mut tumours (p < 0.05). Both TDO2 and IL4I1 were highly expressed high-risk tumours, particularly p53mut cases.
Conclusions
Tryptophan-kynurenine enzymes shape the immune landscape of EC in a subtype-specific manner. High IDO1 was linked to favourable outcomes in p53mut and NSMP cases, whereas TDO2 predicted poor prognosis. The CD163:CD8 ratio emerged as an independent marker of poor survival. These findings support therapeutic strategies combining dual IDO1/TDO2 inhibition or targeting the IL4I1– aryl hydrocarbon receptor (AhR) axis to enhance immunotherapy efficacy in EC
Adaptive sensor steering strategy using deep reinforcement learning for dynamic data acquisition in digital twins
This paper introduces a sensor steering methodology based on deep reinforcement learning (DRL) to enhance the predictive accuracy and decision support capabilities of digital twins by optimizing the data acquisition process. Traditional sensor placement techniques are often constrained by one-off optimization strategies, which limit their applicability for online applications requiring continuous informative data assimilation. The proposed approach addresses this limitation by offering an adaptive framework for sensor placement within the digital twin paradigm. The sensor placement problem is formulated as a Markov decision process (MDP), enabling the training and deployment of an agent capable of dynamically repositioning sensors in response to the evolving conditions of the physical structure as represented by the digital twin. This ensures that the digital twin maintains a highly representative and reliable connection to its physical counterpart. The proposed framework is validated through a series of comprehensive case studies involving a cantilever plate structure subjected to diverse conditions, including healthy and damaged conditions. The results demonstrate the capability of the DRL agent to adaptively reposition sensors, improving the quality of data acquisition and hence enhancing the overall accuracy of digital twins