High-frequency (HF) skywave propagation exploits ionospheric reflection for beyond-line-of-sight transmission, making time-difference-of-arrival (TDOA)-based geolocation a primary technique for localizing non-cooperative HF emitters. However, reliable TDOA estimation remains challenging due to time-varying ionospheric conditions, wideband multipath dispersion, and low signal-to-noise ratio (SNR). This paper proposes an integrated framework coupling realistic channel synthesis, deep learning-based TDOA estimation, and convex optimization-based localization. Three contributions are made. First, an improved wideband ionospheric channel model is constructed by integrating the International Reference Ionosphere (IRI) with region-specific calibration and a stochastic perturbation module, yielding time-varying multipath responses for physics-consistent waveform generation. Second, a convolutional neural network (CNN)-based TDOA estimator is designed to jointly exploit time-domain complex-baseband in-phase/quadrature (I/Q) waveforms, multi-weight generalized cross-correlation (GCC) feature maps, and channel-state information (CSI) within a unified regression network, achieving robust delay estimation under severe noise and multipath conditions. Third, the geolocation problem is formulated as a bias-regularized constrained least-squares problem with unknown ionospheric excess-delay surrogates, and a semidefinite programming (SDP) relaxation is derived to yield a tractable solution without prescribing a fixed virtual reflection height. Simulations show that the proposed estimator consistently outperforms competing algorithms across a wide SNR range and narrows the gap to the Cramér-Rao lower bound (CRLB) at high SNR. On field-recorded signals, the estimator reduces the mean absolute TDOA deviation by 51% relative to GCC with phase transform (GCC-PHAT), and the end-to-end pipeline achieves a mean geolocation error of 19.67 km across 100 field segments, outperforming all compared baselines.
Out-of-hospital cardiac arrest remains a major cause of preventable death. Rapid defibrillation is essential, yet access to automated external defibrillators is often delayed due to poor visibility and limited availability. Real-time geolocation platforms have emerged to improve access and shorten response times. A scoping review was conducted following Joanna Briggs Institute and PRISMA-ScR guidelines. Four databases: PubMed, Scopus, Embase, and Web of Science were searched from inception to April 2025. Studies reporting on real-time geolocation platforms designed to support early defibrillation in community or prehospital settings were included. Fourteen studies from seven countries were included. Most systems were smartphone-based or web applications integrated with emergency medical services. These platforms demonstrated potential to optimize early response: citizen responders arrived before emergency teams in 13-42% of cases, performed cardiopulmonary resuscitation in up to 69%, and delivered defibrillation in nearly 50%. Reported survival ranged from 8% to 88%, and restoration of spontaneous circulation occurred in approximately 30-39% of cases. However, inadequate device visibility (67%), restricted access (36%), and limited geographic coverage remained major operational barriers. Geolocation-based systems may improve community response to cardiac arrest and enhance survival through earlier defibrillation. Rigorous prospective studies are required to evaluate their long-term impact and determine their applicability across different emergency care systems.
Pollen is a robust and widespread substance that captures a historical snapshot of a specific time and place, and it can be used to track movements through space by examining the pollen deposited on various objects. Palynology, the study of pollen, is used across fields such as conservation, natural history, and forensics, where it is particularly useful for tracing the origin and movement of objects. However, pollen has remained underutilized due to the difficulty of distinguishing many pollen taxa beyond the family level and limited pollen reference material to support location predictions. With recent developments in pollen DNA metabarcoding these issues have been rectified, but much of the available pollen data are primarily from wind-pollinated species, which are widespread and less informative of specific sample locations. Bee-collected pollen presents an untapped resource in training predictive models to geolocate sample origin. Here we compiled bee-collected pollen DNA sequence relative abundance data from three projects in the western U.S. and assessed the accuracy of supervised machine learning models to predict the location of sample origin based solely on pollen assemblage, without the need of incorporating additional data. Random Forest and k-Nearest Neighbors models yielded high accuracy across all projects. We also found that models trained on taxonomically clustered pollen assigned sequence variants (ASVs) performed slightly better than those trained on raw sequence data, but the difference was minor, indicating that models trained on raw sequence data can reliably predict location and avoid the time-consuming taxonomic assignment process. Our results demonstrate the utility of repurposing bee-collected pollen for geolocation and provide a framework for employing supervised machine learning in future geolocation efforts. Bee-collected pollen metabarcoding data was used to accurately predict sample originRandom Forest and k-Nearest Neighbors algorithms were most accurate with lowest errorTaxonomically-classified and raw DNA sequence data training sets performed comparably.
The persistent transmission of circulating vaccine-derived poliovirus type 2 (cVDPV2) in Northern Nigeria, especially in the Axis of Intractable Transmission (AIT) states, Kebbi, Sokoto, Katsina, and Zamfara, despite national certification of wild poliovirus interruption, is strongly associated with fragmented settlement visibility, population denominator inaccuracies, and uneven implementation of Supplementary Immunization Activities (SIAs). This study examines whether these geospatial investments as broader ecosystem of interventions improve immunization reach and operational quality during cVDPV2 SIAs. A retrospective operational analysis was conducted using data from nine consecutive SIAs (April 2024-June 2025). We extracted state-level indicators from the Master List of Settlements (MLoS) geospatial records and campaign data, which included (1) geospatial completeness, defined as the proportion of settlements with validated coordinates; (2) settlement tracking coverage; (3) MLoS-derived population targets for children; and (4) number of children vaccinated. Descriptive statistics, bivariate exploratory relationships, and temporal learning analysis were performed to evaluate how geospatial status influenced tracking outcomes. Geospatial completeness ranged from 92.1% (Zamfara) to 100% (Kebbi) and was associated with higher, more stable tracking coverage. Kebbi, with complete geolocation of settlements, demonstrated consistently strong tracking performance (mean 93%). Settlement tracking correlated positively with vaccination outputs in Kebbi and Sokoto but demonstrated saturation effects in Katsina and diminishing returns in Zamfara. Operational learning was observed in Sokoto, Katsina, and Zamfara, each demonstrating 8 to 10 percentage point improvements in tracking coverage across rounds. In addition to other contributory factors, settlement-level geolocation improves the quality and reach of cVDPV2 SIAs by strengthening visibility of target populations and promoting more reliable tracking. The MLoS updates and tracking of settlements and vaccination teams demonstrated transformative potential in optimizing vaccination campaigns by improving equity, accountability, and efficiency. Its deployment offers valuable lessons for enhancing immunization efforts in resource-constrained and security-compromised settings, emphasizing the importance of strategic integration and sustained investments in public health innovation.
Large earthquakes commonly generate surface rupture accompanied by both localized on-fault slip and spatially distributed off-fault deformation. Capturing both components is essential for understanding rupture processes and improving earthquake hazard assessment, yet field mapping alone often fails to fully document diffuse deformation. Here we evaluate the applicability of high-resolution Korea Multi-Purpose Satellite (KOMPSAT)-3 and -3A optical imagery for mapping near-field co-seismic deformation using sub-pixel optical image correlation (OIC), through two case-study areas affected by the 6 February 2023 Kahramanmaraş, Türkiye, earthquake sequence. We processed pre- and post-event stereo-mode KOMPSAT imagery using a MicMac-based workflow to generate orthorectified products and displacement fields, and compared the results with published Sentinel-2 OIC products and independent airborne Light Detection and Ranging (LiDAR) measurements. In the Hatay Airport area, KOMPSAT-3/3A OIC recovered a displacement pattern consistent with Sentinel-2, indicating ~5 m of relative motion across the fault, while the ~1 m effective spatial resolution enabled identification of localized infrastructure offsets (runway displacement) that were not detectable in 10 m Sentinel-2 imagery. In the Elbistan near-epicenter area, KOMPSAT-3/3A OIC resolved block motions of ~6 m and ~2 m in opposing directions. Swath profile analysis indicates an average on-fault slip of 6.8 m, whereas the total slip including distributed deformation reaches 9.3 m, implying that approximately 27% of the deformation is accommodated off-fault. Airborne LiDAR mapping provides an independent benchmark, with on-fault net slip of ~6.13 m and horizontal slip of 5.57 ± 1.40 m, consistent with the KOMPSAT-derived on-fault estimates and supporting the quantitative validity of the OIC results. However, the rupture geometry inferred from OIC is simpler than LiDAR-derived mapping, and absolute geolocation uncertainty remains a limiting factor with a post-correction Root Mean Square Error (RMSE) of 10.25 m and Circular Error with 90% Confidence (CE90) of 11.34 m, requiring cautious interpretation of absolute displacement magnitudes. Overall, our results demonstrate that KOMPSAT-3/3A imagery can serve as an effective resource for rapid rupture mapping and quantifying both on-fault and distributed deformation, while highlighting key requirements for improving geolocation control and integrating complementary datasets for robust three-dimensional deformation assessment.
Current tuberculosis (TB) screening tools, such as the WHO four-symptom screen (W4SS), lack sufficient sensitivity and specificity for effective community-based active case finding, contributing to both missed diagnoses and unnecessary diagnostic evaluations. This study aimed to develop and validate a machine learning (ML) model to improve TB risk prediction among persons aged ≥15 years in community settings of Zambia and South Africa. A large, harmonized dataset was created from four community-based TB prevalence surveys in South Africa and Zambia (N=169,813), restricted to individuals not under treatment at the time of survey. A binary reference outcome was defined based on available microbiological and radiographic data, grouping individuals as either 'Possible TB' or 'Unlikely TB'. An XGBoost model was trained on 80% (N=135,854) of the data using demographic, clinical, and socio-economic variables, and model interpretability was assessed using SHapley Additive exPlanations (SHAP) values. Internal validation was performed using a 20% hold-out test set (N=33,959). Model performance was assessed using discrimination, calibration, and clinical utility measures compared to the W4SS and against WHO's 2025 Target Product Profile (TPP) for a tool in a two-step screening algorithm. Overall, 16,413 (9.7%) of individuals were labelled as 'Possible TB'. On the test set, the XGBoost model yielded an area under the curve (AUC) of 79.7% (95% CI: 78.7, 80.7), outperforming the W4SS (AUC 57.0%; 95% CI: 56.1, 57.8). The XGBoost model achieved 81.5% sensitivity (95% CI: 77.6, 84.9) at a 60% specificity threshold. This exceeded the W4SS, which achieved only 38.2% sensitivity (95% CI: 36.5, 39.9) on the same dataset. SHAP analysis identified age, previous TB treatment, times treated for TB and unemployment as the primary contributors to risk. The ML XGBoost model shows promise as a screening tool to support community-based active case finding activities prior to diagnostic testing. However, as performance remained below TPP targets, and adding variables, e.g. on geolocation, could be considered. The study was not registered.
Identifying and preserving biological diversity is fundamental for the conservation of wild populations. The Atlantic bluefin tuna (Thunnus thynnus, ABT) is an apex predator and vital species to the pelagic ecosystems of the North Atlantic Ocean, with populations now rebounding from decades of overfishing due to strict enforcement of conservation measures. Here, we combine high-resolution whole-genome sequencing data with spatial data from electronic tagging to improve our understanding of population structure in ABT. We analyzed 82 whole-genome sequences obtained from mature fish tracked to geographically distinct spawning grounds, as well as larvae representing the two recognized stocks (western and eastern) of ABT. We obtained 11,181,223 single-nucleotide polymorphisms (SNPs) and integrated these genomic data with 12,974 total geolocation days of adult ABT (mean individual deployment length: 271 ± 110.4 days). This extensive dataset of electronic tracks enables spatial assignment of individuals to their respective spawning grounds and the first whole-genome comparison of migratory phenotypes. Both neutral and adaptive SNP markers reflect the same genomic population structure as the spatial movement patterns, likely maintained by natal philopatry, and we highlight candidate genes with potentially adaptive roles. Our analyses show that the two populations diverged ∼27,000 years ago, overlapping with the Last Glacial Maximum, and we suggest that oceanographic variation of the spawning grounds has contributed to shaping present-day bluefin tuna genomic diversity. Overall, these results improve our understanding of adaptive variation in bluefin tuna, which will be important for management decisions.
The location of postgraduate training is considered a key determinant for the future practice location of primary care physicians. This study evaluates the effectiveness of the GP training program in Czechia, focusing on the spatial and temporal links between the place of training and subsequent career entry. We conducted a longitudinal analysis of contractual data from the General Health Insurance Company (GHIC). While GHIC insures approximately 60% of the Czech population, it maintains contracts with virtually all GP practices providing standard care in Czechia, ensuring near-complete coverage of the primary care provider network. We tracked 661 GP trainees who completed their training between 2009 and 2022. Career paths were analyzed using geolocation mapping at four administrative levels (from municipality to region) alongside data on age, sex, and full-time equivalent (FTE) workloads. Entry into independent practice is a gradual process: only 55% of doctors established a contract within the first year (44% specifically in GP), increasing to 83% after seven years (79% in GP). A significant gender gap was observed: men entered practice more rapidly and with higher FTEs, while women (representing 73% of the cohort) showed a slower, steadier increase in practice involvement. Geographic retention at the regional level (NUTS 3) reached 65% for part-time and 49% for full-time positions after seven years. Retention at the municipal level (LAU 2) was lower (40% and 28% respectively), often linked to trainees taking over established practices from their trainers. The study confirms the ongoing feminization of primary care and a growing preference for flexible workloads among young GPs. While the place of training significantly influences future practice location, the transition to independent practice is not immediate. These findings suggest that healthcare workforce planning should account for the specific career trajectories of female physicians and the importance of regional training capacities to ensure long-term stability in primary care.
Beach litter remains a persistent threat to coastal ecosystems, with far-reaching ecological, social, and economic consequences. Although official statistics provide essential information, they frequently overlook the numerous voluntary beach clean-up initiatives led by local associations, schools, companies, and informal groups. This study evaluates the potential of social media as a complementary data source for documenting such activities. Spain was selected as a case study due to its extensive coastline and strong reliance on tourism. Through keyword-based searches and geolocation techniques applied to posts on X and Instagram, we identified 487 beach-cleaning events in 2024, of which 458 (94%) were absent from official registries. These data reveal spatial and temporal patterns not captured by traditional monitoring systems, demonstrating the added value of user-generated content for coastal stewardship assessment. We also estimated the economic contribution of these initiatives from €1.13 to €2.86 million, depending on the approach, representing approximately 0.71-1.80% of Spain's annual public expenditure on beach litter removal. Beyond their measurable economic and ecological value, beach clean-ups promote social participation, environmental awareness, and engagement in marine conservation. Overall, our findings show that social media monitoring can effectively complement official statistics by providing a more comprehensive understanding of the scope and contribution of citizen-led efforts in coastal management and marine-litter mitigation.
Insider-originated breaches remain the predominant vector of data compromise in healthcare. Traditional role-based access control (RBAC) offers limited responsiveness to dynamic clinical workflows. This paper presents an adaptive access governance model integrating RBAC, attribute-based, and risk-based authorisation with continuous behavioural anomaly detection. Contextual factors including temporal patterns, device provenance, geolocation, and record sensitivity are incorporated into real-time access logic supported by dynamic risk scoring. Validation in simulated and operational environments demonstrates reduced inappropriate access events and enhanced alignment with GDPR, HIPAA, and ISO/IEC 27001.
Whether living environment may influence outcome of stroke survivors remains to be elucidated. This registry-based cohort study aimed to assess the relationship between urban greenness around the residence and one-year death or recurrence after a first-ever ischaemic stroke. Patients with a first-ever ischaemic stroke who directly returned home were identified from the population-based registry of Dijon, France. For each patient, after geolocation of residential building, two greenness indices were calculated: the distance by road and pedestrian networks to the nearest public green space, and the area of green spaces within radii of 100 and 400 metres. Atmospheric NO2 and PM10 outdoor concentrations around the residence and deprivation index were assessed. During the 2005-2008 study period, 360 patients were identified and included (median age: 75 years-old (IQR: 63-83), 56% women). Fifteen died and 17 had recurrent stroke during the one year of follow-up. In adjusted models, the distance between public green spaces and patients' residence was associated with stroke recurrence or death (HR = 1.26, 95% CI: 1.08-1.48, P < 0.01, for each 100 metre section of city network). In age-stratified analysis, this association remained significant only in patients aged 65-79 years (HR: 1.37, 95% CI: 1.10-1.71, P < 0.01). When considering separately stroke recurrence and death, this association remained significant for recurrence (HR = 1.30, 95% CI: 1.07-1.58, P < 0.01) but not for death (HR = 1.17, 95% CI: 0.89-1.52). This study highlighted a beneficial influence of greenness on post-stroke recurrence in an urban area. These results indicate that urban planning policy could impact secondary prevention.
This study examined the endorsement of electronic screening and brief intervention (e-SBI) features among men who have sex with men (MSM) who use substances and live in rural Southern U.S. counties. Additionally, demographic, care access, and substance use correlates of endorsed features were assessed. Participants (N = 412) completed an online cross-sectional survey. Descriptive statistics and split logistic regression models were employed. Over half of participants endorsed features related to substance misuse screenings, a list of local behavioral health professionals, substance misuse prevention information, and virtual communication with a behavioral health professional. Young adults were less likely, whereas racial minority participants were more likely, to support multiple e-SBI features. Employed participants had lower odds of preferring substance use screenings, while health insured participants had higher odds of endorsing in- and outpatient program listings. Participants who reported alcohol and stimulant use were more likely to select geolocation-based notifications, while participants reporting depressants and dissociative use were more likely to endorse listings of and talking with local behavioral health professionals. Findings highlight the heterogeneity in e-SBI feature preferences across demographic and substance use profiles, suggesting that user-informed design approaches may enhance the acceptability and uptake of e-SBIs among rural MSM.
Cultural contexts, such as whether one's immediate environment is ethnoracially congruent, are known to influence emotional expression, emotional experience, motivation, and social behavior in healthy individuals. However, it is unclear whether such cultural factors play a role in state exacerbations in negative symptoms that occur in schizophrenia (SZ). The current study combined GPS data, environmental geocoding, and ecological momentary assessment (EMA) to test the hypothesis that ethnoracial incongruence encountered in daily-life situations predicts state increases in negative symptoms in SZ. Participants included outpatients with SZ (n = 37) and healthy controls (CN: n = 41) with marginalized ethnoracial identities who completed EMA and passive digital phenotyping recordings. Geolocation was used to pair participant GPS location at the time of completing EMA symptom surveys with geocoded measures of that location's ethnoracial density based on government census records. Ethnoracial congruence was determined in relation to the match between a participant's identified ethnoracial identity and the ethnoracial density of their location at the time of EMA survey. Results indicated that ethnoracially incongruent contexts were associated with state increases in negative symptoms in individuals with SZ, but not CN. These findings suggest that interactions between one's own ethnoracial identity and the ethnoracial context of the current environment contributes to negative symptom exacerbations in SZ. Identity factors are not typically considered in the assessment and treatment of negative symptoms in SZ, but it would be beneficial to do so.
Waterpipe tobacco smoking remains a popular social activity among young adults in the United States. This study aims to understand sentiment toward waterpipe on social media in the United States by analyzing Twitter/X data. Using keywords ("hookah," "waterpipe," "shisha"), we collected US tweets posted between March 2021 and March 2023. Commercial content (eg, containing "sale," "discount," "$") was filtered out, yielding 299 544 non-commercial tweets. A random sample of 2300 tweets was manually coded for sentiment (positive/negative/neutral) and for whether the author might use waterpipe, which were used to fine-tune a deep-learning model (Llama-2). We applied BERTopic modeling to identify main themes. Overall, tweets with a positive sentiment (57.0%, 170 597/299 544) were higher than those with a negative sentiment (16.7%, 50 196/299 544). Among those Twitter users who might use waterpipe, 82.0% of their tweets showed a positive sentiment toward waterpipe while only 7.5% of the tweets had a negative sentiment. In contrast, among those who might not use waterpipe, 29.6% of their posts showed a positive sentiment and 27.0% of the posts with a negative sentiment. Main themes identified from positive tweets included excitement about waterpipe lounges, enjoyment of specific flavors, and social desires for waterpipe use. Negative tweets focused on health concerns of waterpipe, discomfort from waterpipe smoking, criticism of promotional content, and negative social sentiments. Our results provide preliminary insights into how waterpipe smoking was perceived and discussed among Twitter users in the United States, which could help with future targeted social media-based public health intervention campaigns. By applying fine-tuned Llama-2 language models and BERTopic modeling, this study showed how the public perceived waterpipe on Twitter, which varied depending on the user status (whether they used waterpipe or not), time of day or day of week, and geolocation. These findings offer some insights into the waterpipe discourse on Twitter. More importantly, our understanding of different sentiments toward waterpipe will provide useful guidance for designing effective health communication to reduce or prevent waterpipe use.
Smartphones and wearables are low-burden tools for assessing real-time mood and behavior. Although these methods have been used with adolescents for behavioral tracking (e.g., activity, sleep), less is known about longer-term use (beyond one week) with adolescents with depression and about mobile sensing for monitoring mood for any adolescent population. This study examined acceptability and feasibility of a one-month EMA, actigraphy, and mobile sensing protocol for adolescents with elevated depressive symptoms. Adolescents aged 12 to 18 (N = 69; Mage = 15.46; 67% assigned female at birth; 42% White; 71% Hispanic or Latine; 38% sexual minority) completed EMA surveys on depressive symptoms, processes, and affect multiple times daily via a smartphone app that also collected passive sensor data (e.g., motion, geolocation). An actigraph measured physical activity and sleep. A feedback interview assessed protocol acceptability. Most participants (91%) completed all components, were willing to participate again (91%), and would recommend participation to peers (93%). EMA response rates improved (mean completion 57% to 66%) after shifting to a semi-personalized schedule with extended response windows. Actigraph wear time was high (> 70%) despite device-related issues. Sensor data availability varied by operating system, and privacy concerns influenced participation. Adherence was correlated within and between modalities, suggesting that individual compliance played a central role in consistent engagement. Findings support the feasibility and acceptability of smartphone and wearable methods for capturing real-world mood and behavior in adolescents, however careful attention to design, engagement, and ethical considerations remains essential.
Walking activity can improve the prognosis of people with chronic obstructive pulmonary disease (COPD), but might worsen the harmful effects of pollution if practised in polluted environments. We aimed to determine the combined effect of walking activity and air pollution on daily respiratory symptoms in people with COPD. This multicentre panel study assessed 105 people with COPD from Catalonia (Spain) over two 7 days periods. Daily walking activity (walking duration, step count) was measured via activity monitors. Daily air pollutant concentration (particulate matter with an aerodynamic diameter <2.5 µm (PM2.5), black carbon (BC), nitrogen dioxide (NO2)) was estimated by combining geolocation data with land use regression models. Dyspnoea, cough, expectoration and wheezing were recorded in diaries every evening (0-10 scale). We estimated the individual association of daily walking activity and air pollution and their interaction on symptoms using multivariable mixed-effects negative binomial models. Participants (20% female) were mean±SD 68±7 years old and had a forced expiratory volume in 1 s of 53±23% predicted. Walking duration was associated with higher levels of cough and expectoration. All pollutants were associated with higher cough and BC also with higher expectoration, dyspnoea and wheezing. After including an interaction term, more walking duration was associated with higher cough and expectoration only on days with high BC concentrations (p for interaction <0.05). No interaction was observed between walking duration and PM2.5 or NO2. Results were similar for step count. Daily walking activity may worsen cough and expectoration in people with COPD when BC concentrations are high. Actions to ensure accessible low-traffic spaces to walk are paramount.
Health research has shifted from a disease-centered approach towards emphasizing functioning and more specific lived health. Lived health, the actual performance of daily activities in one's environment, has nevertheless received limited attention, and its assessment remains methodologically challenging. Ecological Momentary Assessment (EMA), a real-time method capturing behaviors, emotions, and context in natural settings, holds promise in this regard. Although EMA research is increasing, insight into its use for studying daily activities is still limited. To address this gap, this study systematically maps EMA applications across diverse health and disability populations to better understand lived health, defined as actual engagement in daily activities. A bibliometric analysis was conducted using a literature search on Web of Science with keywords related to EMA combined with daily activity terms, yielding 3,692 English-language articles. Publications were classified according to general characteristics, distribution of disability and health populations, actual engagement in daily activities following the person-environment-occupational model (PEO-model), and interaction analyses combining the last two analyses. The results show that mental disorders dominate the EMA research on daily activities, representing 75% of the dataset, which has significantly shaped the overall research landscape. Moreover, while personal factors are frequently highlighted, occupational and environmental dimensions remain underrepresented. These findings suggest that future EMA research should better integrate aspects of person, occupation, and environment, for instance by using tools such as geolocation and passive sensing to capture daily functioning more holistically. Expanding research beyond mental health and increasing secondary analyses will further strengthen the relevance and impact of EMA on health research.
Real-time fire detection and precise geographic localization using unmanned aerial vehicles (UAVs) are critical for early forest-fire warning. However, existing approaches face a fundamental dilemma: achieving high detection accuracy requires complex deep learning models with prohibitive computational costs, while lightweight models sacrifice localization precision due to the lack of spatial priors. This study proposes BGC-LiteNet, an end-to-end framework that integrates the BeiDou Grid Code (BGC)-China's national spatial reference standard-directly into neural network feature learning. A learnable geographic embedding module encodes pixel-grid correspondences at the input stage, enabling simultaneous detection and localization without external GIS post-processing. To achieve millisecond-level inference on resource-constrained UAV platforms, we develop a latency-aware lightweight neural architecture search (L-NAS) that jointly optimizes detection accuracy and hardware latency. Experimental results on a multi-scenario UAV dataset demonstrate that BGC-LiteNet achieves 88.9% mean average precision (mAP) and 92.4% geolocation accuracy with only 0.87 M parameters and 38.2 ms latency on embedded platforms. The model maintains robust performance under challenging conditions including low illumination (mAP 72.3%), dense smoke (mAP 72.6%), and small fire points (recall 86.7%). By embedding structured geographic priors and optimizing model latency simultaneously, BGC-LiteNet establishes a new paradigm for spatiotemporal intelligent edge computing in disaster prevention applications.
Suicide represents a major public health issue influenced by a complex interplay of individual, social, and environmental factors. While suicidological research traditionally focuses on psychological and clinical aspects, spatial dimensions of suicidal behavior remain less explored, particularly in Central Europe. This study presents a geospatial analysis of completed suicides in the Olomouc and Zlín Regions of the Czech Republic between 2018 and 2022, using a unique dataset derived from forensic autopsy records. The primary goal was to demonstrate the potential of spatial processing and interdisciplinary collaboration between forensic medicine and geoinformatics in identifying contextual patterns of suicidal behavior. A dataset of 585 completed suicides was compiled from forensic autopsy reports, including detailed individual-level characteristics and geolocation data. Spatial analyses were conducted using GIS software, integrating additional layers such as land use, demographic and socioeconomic indicators, and quality of life indices. Both point-level and aggregated data (municipality and administrative district levels) were used to explore correlations and spatial variability. The study confirmed known trends - such as male predominance and the role of alcohol - and identified new spatial relationships, including a negative correlation between blood alcohol concentration and latitude, and between age and distance from residence to suicide location. Spatial analyses were conducted at multiple levels of aggregation and combined with selected socioeconomic and environmental indicators, including quality of life, urban-rural context, and foreclosure (enforcement) rates. The results demonstrate that spatial patterns and correlations between suicide rates and area-based characteristics vary depending on the spatial scale of analysis. The findings illustrate the potential of interdisciplinary collaboration between forensic medicine and geoinformatics and provide an exploratory basis for further research and more context-sensitive suicide prevention approaches.
The ongoing drug poisoning crisis continues to cause significant mortality, with a disproportionate number of overdose deaths occurring when individuals use drugs alone. While supervised consumption sites (SCS) have proven effective in reducing overdose fatalities, their impact is limited by geographic, social, and systemic barriers. In response, overdose response technologies have emerged to expand access to life-saving interventions beyond the reach of traditional harm reduction infrastructure. Overdose response technologies (e.g., National Overdose Response Service (NORS)) and applications (e.g., Lifeguard App, UnityPhilly) offer real-time monitoring during solitary substance use. Hotlines provide peer-operated support and activate emergency responses if a caller becomes unresponsive, while apps use timers and geolocation to trigger automatic emergency services dispatch. Despite promising early outcomes, these services operate in a fragmented policy landscape without formalized regulatory guidance or implementation best practices. Preliminary data show that services like NORS have successfully prevented overdose deaths; however, published outcomes for most services remain limited. Key areas of priority for standards include the following: ensuring privacy for service, balancing data usage for quality improvement and research, building capacity to further equity of access to healthcare and harm reduction using the virtual platform, standardizing overdose response, and providing appropriate education around the efficacy of services. To enhance the effectiveness and sustainability of overdose response technologies, a comprehensive policy or standards framework is needed. This includes guidance on data privacy, service equity, public education, capacity-building, and outcome evaluation, laying the groundwork for safer, scalable, and more accessible overdose prevention interventions. RéSUMé: CONTEXTE: La crise actuelle liée à l'intoxication médicamenteuse continue d'entraîner une mortalité importante, avec un nombre disproportionné de décès par surdose chez les personnes solitaires qui consomment des drogues. Si les sites de consommation supervisée (SCS) se sont avérés efficaces pour réduire le nombre de décès par surdose, leur impact est limité par des obstacles géographiques, sociaux et systémiques. En réponse à cela, des technologies d'intervention en cas de surdose ont vu le jour afin d'élargir l'accès à des interventions vitales au-delà de la portée des infrastructures traditionnelles de réduction des risques. INTERVENTION: Les technologies d'intervention en cas de surdose (par exemple, le National Overdose Response Service [NORS]) et les applications (par exemple, Lifeguard App, UnityPhilly) offrent une surveillance en temps réel pendant la consommation solitaire de substances. Les lignes d'assistance téléphonique fournissent un soutien assuré par des pairs et activent les services d'urgence si l'appelant ne répond plus, tandis que les applications utilisent des minuteries et la géolocalisation pour déclencher l'envoi automatique des services d'urgence. Malgré des résultats prometteurs, ces services fonctionnent dans un contexte politique fragmenté, sans directives réglementaires formelles ni de bonnes pratiques de mise en œuvre. RéSULTATS: Les données préliminaires montrent que des services tels que le NORS ont permis de prévenir avec succès des décès par surdose; toutefois, les résultats publiés pour la plupart des services restent limités. Les principaux domaines prioritaires pour les normes sont les suivants: garantir la confidentialité du service; trouver un équilibre entre l'utilisation des données à des fins d'amélioration de la qualité et de recherche; renforcer les capacités afin de favoriser l'équité dans l'accès aux soins de santé et à la réduction des risques à l'aide de la plateforme virtuelle; standardiser les interventions en cas de surdose; et fournir une éducation appropriée sur l'efficacité des services. IMPLICATIONS: Afin d'améliorer l'efficacité et la durabilité des technologies d'intervention en cas de surdose, un cadre politique ou normatif complet est nécessaire. Celui-ci doit inclure des orientations sur la confidentialité des données, l'équité des services, l'éducation du public, le renforcement des capacités et l'évaluation des résultats, afin de jeter les bases des interventions de prévention des surdoses plus sûres, plus évolutives et plus accessibles.