Air Quality Atmosphere and Health最新文献

{"title":"Association between mean temperature and hospital admissions for acute ischemic stroke: a time-series study","authors":"Xiaoxiao Song,&nbsp;Wei Chen,&nbsp;Xiaolan Huang,&nbsp;Zhiling Cai,&nbsp;Xiaodong Wen,&nbsp;Jialei Yang,&nbsp;Lian Gu","doi":"10.1007/s11869-026-02102-5","DOIUrl":"10.1007/s11869-026-02102-5","url":null,"abstract":"<div><p>Ambient temperature has been associated with cardiovascular and cerebrovascular outcomes, but its short-term relationship with acute ischemic stroke (AIS) admissions remains heterogeneous across regions. This study evaluated the association between mean temperature (MT) and AIS admissions in a subtropical city in China. Daily AIS admissions from July 1, 2017, to June 30, 2020, were collected from a tertiary hospital in Nanning, China. A quasi-Poisson generalized linear model combined with a distributed lag nonlinear model was used to assess the nonlinear and lagged associations between MT and AIS admissions over lag days 0–7. Sensitivity analyses were conducted by varying temporal adjustment, air-pollutant adjustment, maximum lag periods, excluding 2020, and adjusting for the Spring Festival window. Among 2,382 AIS admissions, the mean daily number of admissions was 2.17. Compared with the median MT of 23.30 °C, higher MT (29.60 °C) was associated with increased AIS admissions over early cumulative lag periods, with relative risks of 1.347 (95% CI: 1.058–1.714) for lag0-1 and 1.396 (95% CI: 1.085–1.797) for lag0-2. The association attenuated over longer lag periods. Lower MT (10.50 °C) showed a lower same-day estimate but less consistent cumulative associations. The main findings were generally robust in sensitivity analyses. Higher MT was associated with a short-term increase in AIS admissions in this subtropical setting, with the association concentrated within the early lag period. The low-MT association was less consistent and requires cautious interpretation.</p></div>","PeriodicalId":49109,"journal":{"name":"Air Quality Atmosphere and Health","volume":"19 10","pages":""},"PeriodicalIF":3.1,"publicationDate":"2026-09-24","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149035286","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"环境科学与生态学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"A retrieval-augmented historical knowledge-base framework with residual correction for station-level ozone forecasting across urban agglomerations","authors":"Mingkun Zhu,&nbsp;Xiaoxia Han,&nbsp;Jinde Wu,&nbsp;Haonan Zhu,&nbsp;Wenxin Chai","doi":"10.1007/s11869-026-02105-2","DOIUrl":"10.1007/s11869-026-02105-2","url":null,"abstract":"<div><p>Accurate station-level ozone forecasting across urban agglomerations remains challenging because ozone evolution is affected by nonlinear pollutant–meteorology interactions, regional transport, local heterogeneity, and interannual distribution shifts. Most data-driven models encode historical patterns implicitly in trained parameters, limiting their ability to explicitly reuse recurring pollution episodes. We propose OzoneKBNet, a retrieval-augmented historical knowledge-base framework that forecasts station-level ozone concentrations over the next 48 h from the preceding 96 h observations of <span>(O_3)</span>, <span>(NO_2)</span>, <span>(PM_{2.5})</span>, temperature, and relative humidity. It retrieves multi-scale cross-station analogs, re-ranks them by representation similarity and local-trend consistency, and uses their future trajectories as bounded, reliability-aware corrections to a lightweight LSTM forecast. To prevent temporal leakage, observations from 2023 are used for normalization, retrieval-encoder pretraining, and knowledge-base construction, data from 2024 for supervised training, and fixed samples from 2025 for testing. Across seven Chinese urban agglomerations, OzoneKBNet achieves macro-average MAE and RMSE values of 20.06 and 24.45 <span>(mu textrm{g},textrm{m}^{-3})</span>, respectively, and the lowest regional MAE in six of the seven regions. Relative to station-specific LSTM, it reduces MAE and RMSE by 1.6% and 1.2%, respectively. These results indicate that explicit regional historical evidence can complement parametric forecasting under the evaluated cross-year setting and may support short-term ozone-pollution episode warning.</p></div>","PeriodicalId":49109,"journal":{"name":"Air Quality Atmosphere and Health","volume":"19 10","pages":""},"PeriodicalIF":3.1,"publicationDate":"2026-09-24","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149035287","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"环境科学与生态学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Association of air pollutants and PM2.5 components with pediatric outpatient visits for acute respiratory infections in Baoshan District, Shanghai","authors":"Duo Wang,&nbsp;Chunyang Dong,&nbsp;Cheng Yang,&nbsp;Yewen Shi,&nbsp;Feier Chen,&nbsp;Zheng Wu,&nbsp;Mengshuang Liu,&nbsp;Xianbiao Shen,&nbsp;Tian Chen,&nbsp;Jianghua Zhang","doi":"10.1007/s11869-026-02106-1","DOIUrl":"10.1007/s11869-026-02106-1","url":null,"abstract":"<div><p>Although ambient air pollution has been associated with increased outpatient visits for respiratory diseases, the health effects of PM_2.5 chemical constituents on pediatric outpatient visits for acute upper respiratory infections (AURI) and acute lower respiratory infections (ALRI) remain insufficiently characterized. This time-series study, conducted in Shanghai from 2018 to 2019, employed generalized additive models to examine the associations of air pollutants (SO₂, NO₂, CO, O₃, PM₁₀, PM_2.5) and PM_2.5 components with outpatient visits for total respiratory diseases (TPRD), AURI, and ALRI. Sensitivity analyses were conducted to assess the robustness of the findings. Statistically significant positive associations were observed between NO₂ and CO and outpatient visits for TPRD and AURI. PM₁₀ was significantly associated with outpatient visits for both AURI and ALRI, whereas PM_2.5 was significantly associated only with AURI. In analyses based on a limited, non‑continuous sampling framework comprising 174 sampling days, the water-soluble ions NO₃⁻ and NH₄⁺ were significantly associated with outpatient visits for TPRD and ALRI, while SO₄²⁻ was significantly associated with ALRI. No statistically significant associations were detected for the measured metal/metalloid components. Overall, these findings support the consideration of PM_2.5 chemical composition, in addition to PM_2.5 mass concentrations and gaseous pollutants, when evaluating the short-term respiratory health effects of ambient air pollution in children.</p></div>","PeriodicalId":49109,"journal":{"name":"Air Quality Atmosphere and Health","volume":"19 10","pages":""},"PeriodicalIF":3.1,"publicationDate":"2026-09-23","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149034507","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"环境科学与生态学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Indoor exposure to microplastics in dust: polymeric composition and health risks in a tropical savanna climatic region","authors":"Raja Sibhi M,&nbsp;Lekshmi Mohan V,&nbsp;Amit Passi","doi":"10.1007/s11869-026-02107-0","DOIUrl":"10.1007/s11869-026-02107-0","url":null,"abstract":"<div><p>Indoor dust bound microplastics (MPs) can be inadvertently inhaled or ingested, especially by children and infants. This study investigated MPs in indoor dust collected from an academic office and two residential dwellings in Tiruchirappalli, South India, representing contrasting urban microenvironments (occupational, traffic and industrial-influenced). Polymer identification using micro-Fourier infrared (µ-FTIR) spectroscopy revealed the dominance of polymers such as, polyethylene terephthalate (PET), nylon, polystyrene (PS), and polyethylene (PE) in indoor dust across all sites. MPs deposition differed significantly among the sites (Kruskal-Wallis, <i>p</i> = 0.027), with highest mean deposition recorded at academic office (Site A; 241 ± 23 particles/m²/day), followed by residence located near busy traffic area (Site C; 151 ± 24 particles/m²/day) and residence located near a small-scale industrial site with lesser traffic (Site B; 54 ± 3 particles/m²/day). The particle size also varied significantly among sites (<i>p</i> &lt; 0.001), with Site A showing significantly smaller MPs (18.9 ± 11.2 μm) than Sites B and C (33–36 μm), likely due to greater occupant activity, lesser frequency of cleaning, and source diversity in the office environment. Site A recorded highest potential polyurethane (PU)- and polyvinyl chloride (PVC)- associated exposure risk (1750 hazard units), and Site C recorded a potential PU-associated hazard ranking (1500 hazard units), while Site B exhibited negligible polymer-associated potential hazard score, indicating that site classification (traffic vs. industrial) does not align with exposure risk. This study emphasizes the need for a comprehensive understanding and characterization of the sources of indoor MPs, as mere count alone cannot determine risk and inform targeted mitigation strategies.</p><h3>Graphical abstract</h3><div><figure><div><div><picture><source><img></source></picture></div></div></figure></div></div>","PeriodicalId":49109,"journal":{"name":"Air Quality Atmosphere and Health","volume":"19 10","pages":""},"PeriodicalIF":3.1,"publicationDate":"2026-09-23","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149034449","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"环境科学与生态学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Temporal variation of outdoor particulate matter concentration in a village of the Qinghai-Tibet Plateau and its correlation with meteorological factors: based on long-term field measurement and machine learning approach","authors":"Yumiao Li,&nbsp;Jialin Chen,&nbsp;Di Wu,&nbsp;Rongjiang Ma,&nbsp;Yuhao Yi,&nbsp;Tao Yu,&nbsp;Mengsi Deng","doi":"10.1007/s11869-026-02094-2","DOIUrl":"10.1007/s11869-026-02094-2","url":null,"abstract":"<div><p>The incomplete combustion of yak dung on the Qinghai-Tibet Plateau (QTP) emitted a substantial amount of PM, causing severe environmental pollution in the villages. However, the temporal variation in pollutant concentrations and the impact of meteorological parameters in the villages of the QTP have not been thoroughly studied. Therefore, this study collected year-round PM concentrations and various meteorological parameters on the QTP. Temporal variation characteristics of PM were analyzed. An integrated analytical framework combining Spearman correlation analysis and machine learning methods was developed to evaluate the correlations between PM_2.5 concentration and meteorological factors. The results showed that the annual mean PM₁₀ and PM_2.5 concentrations were 24.3 ug/m³ and 18.5 ug/m³, the highest monthly mean concentrations of PM₁₀ and PM_2.5 were recorded in December, and the lowest were in July. The intense peaks of PM concentrations occurred in the 8:00-10:00 and 18:00-22:00 periods. The PM concentration peaks in the heating season were much higher than those in the non-heating season. PM_2.5 was the dominant part of the PM with the monthly PM_2.5/PM₁₀ ratio of 68.6%-76.2%. Random Forest (RF) outperformed the Extreme Gradient Boosting (XGBoost) and Light Gradient Boosting Machine (LightGBM) in the aspects of three evaluation metrics. PM_2.5 concentration was negatively correlated with solar radiation, temperature, and wind speed. The outdoor PM_2.5 concentrations during the yak dung stove-operation period in the villages of four cities on the QTP were predicted, which exhibited similar variation trends. The results could help propose future strategies for improving outdoor air quality and assess their environmental-health improvements.</p></div>","PeriodicalId":49109,"journal":{"name":"Air Quality Atmosphere and Health","volume":"19 10","pages":""},"PeriodicalIF":3.1,"publicationDate":"2026-09-23","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149034491","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"环境科学与生态学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Secondhand smoke exposure and depression among never smokers in a prospective cohort study","authors":"Mengkun Li,&nbsp;Changgui Kou,&nbsp;Xijie He,&nbsp;Xinru Guo,&nbsp;Hewanmeng Geng,&nbsp;Zihan Wang,&nbsp;Chenxu Su,&nbsp;Xinyue Zhang,&nbsp;Puheng You,&nbsp;Yan Sun,&nbsp;Buya Batebayier,&nbsp;Pin Han,&nbsp;Wei Bai,&nbsp;Yong Li","doi":"10.1007/s11869-026-02100-7","DOIUrl":"10.1007/s11869-026-02100-7","url":null,"abstract":"<div><p>This study aimed to investigate the association between SHS exposure and depression and to assess whether genetic evidence supports a potential causal relationship in a large cohort of never smokers. This prospective cohort study included 190,403 never smokers from the UK Biobank. SHS exposure was assessed via self-reported hours per week at home and outside home, categorized as exposed (≥ 1 h/week) versus unexposed. Cox proportional hazards models were employed to assess the association between SHS exposure and depression, with dose-response and subgroup analyses further characterizing the relationship. Mendelian randomization (MR) analysis was performed to evaluate genetic evidence supporting a potential causal relationship. During 14.3 years of follow-up, 7,013 incident depression cases occurred. SHS exposure was associated with an increased risk of depression (HR = 1.226, 95% CI: 1.159–1.296). A dose-response relationship was observed, with each additional hour of weekly SHS exposure associated with higher depression risk (HR = 1.012, 95% CI: 1.009–1.016). The association was more pronounced among individuals with high socioeconomic deprivation (<i>P</i> interaction = 0.001). MR analysis using the inverse variance weighted (IVW) method showed a positive genetic association (OR = 1.028, 95% CI: 1.001–1.056). This study provides observational evidence supporting an association between SHS exposure and depression among never smokers, while genetic evidence further supports the possibility of a causal relationship. These findings highlight the importance of reducing involuntary tobacco smoke exposure.</p></div>","PeriodicalId":49109,"journal":{"name":"Air Quality Atmosphere and Health","volume":"19 10","pages":""},"PeriodicalIF":3.1,"publicationDate":"2026-09-18","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148981305","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"环境科学与生态学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Delayed influenza resurgence following high PM2.5 episodes: insights from the 2019 spring anomaly in South Korea","authors":"Min Ju Yeo,&nbsp;Hyunjung Lee,&nbsp;Jinsoo Park,&nbsp;Ja-Ho Koo","doi":"10.1007/s11869-026-02096-0","DOIUrl":"10.1007/s11869-026-02096-0","url":null,"abstract":"<div><p>Most studies linking PM_2.5 to influenza focus on delays of days, leaving uncertain whether severe pollution episodes can precede influenza resurgence weeks later. Using nationwide weekly PM_2.5 observations and laboratory-confirmed influenza surveillance data in South Korea during 2015–2025 (excluding 2020–2022), we applied repeated single-lag and constrained distributed lag models to anomalies calculated by epidemiological week. Analyses covered all weeks and epidemiological weeks (W) 8–20 (W8–20). Sensitivity analyses excluded 2019. No robust lagged association was evident across all weeks. During W8–20, however, estimates increased consistently toward lags 5–6 in both models. The pattern was most consistent for P90 (the 90th percentile of weekly PM_2.5 concentrations across monitoring stations nationwide) and remained significant at lags 5 and 6 after excluding 2019. During W7–9 of 2019, weekly mean PM_2.5 and P90 concentrations averaged 53.3 µg/m³ and 70.1 µg/m³, respectively. Influenza resurged, peaking at W15–16. During W12–16, mean temperature and relative humidity ranked lowest and second lowest, respectively, among the study years. Adjustment for delayed meteorological effects attenuated estimates at later lags. The resurgence was also evident in pediatric surveillance data, with no comparable increase in other respiratory viruses in pediatric hospitalization surveillance. The lag analyses identified an association at later lags during W8–20 rather than throughout the year. Severe PM_2.5 episodes may precede delayed influenza resurgence when followed by conditions favorable for transmission. Air quality information may therefore provide a supplementary environmental signal for intensified influenza surveillance over subsequent weeks.</p></div>","PeriodicalId":49109,"journal":{"name":"Air Quality Atmosphere and Health","volume":"19 9","pages":""},"PeriodicalIF":3.1,"publicationDate":"2026-09-15","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148986488","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"环境科学与生态学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Mortality burden of air pollution in an Iranian industrial city: a 10-year multi-pollutant time-series study","authors":"Behrooz Karimi,&nbsp;Mohsen Farzin,&nbsp;Seyed Hamed Mirhoseini","doi":"10.1007/s11869-026-02098-y","DOIUrl":"10.1007/s11869-026-02098-y","url":null,"abstract":"<div><p>Air pollution remains a major public health concern in highly polluted industrial cities, particularly in low- and middle-income countries. This study quantified the short-term associations between ambient air pollution and cause-specific mortality in Arak, Iran, a representative industrial urban setting with chronically elevated pollutant levels. A 10-year time-series analysis (January 2015–November 2025) was conducted using daily concentrations of PM₂.₅, PM₁₀, SO₂, NO₂, NO, O₃, and CO and daily counts of all-cause, cardiovascular, and respiratory mortality. Generalized linear models with a negative binomial distribution were applied in single- and multi-pollutant frameworks, adjusting for meteorological variables, long-term trends, and temporal confounders. The strongest mortality associations were observed within a lag window of 0–3 days. In single-pollutant models, PM₂.₅ and SO₂ showed significant positive associations with total mortality per interquartile range increase (PM₂.₅ RR = 1.112, 95% CI: 1.082–1.142; SO₂ RR = 1.102, 95% CI: 1.074–1.131), with the largest effects observed for respiratory mortality. Multi-pollutant models confirmed independent associations for PM₂.₅ (RR = 1.086) and SO₂ (RR = 1.092), while the NO₂ association was attenuated after adjustment for co-pollutants—likely reflecting its role as a marker of traffic-related mixtures rather than an independent toxicant. Seasonal analyses revealed stronger PM₂.₅-related mortality risks during summer months, possibly due to enhanced photochemical activity. An estimated 527 excess deaths annually (5.0% of total mortality) were attributable to ambient air pollution exposure, with respiratory mortality showing the highest vulnerability. These findings demonstrate a substantial and preventable mortality burden associated with air pollution in industrial urban environments and highlight the importance of prioritizing emission control strategies targeting particulate matter and sulfur-containing pollutants.</p></div>","PeriodicalId":49109,"journal":{"name":"Air Quality Atmosphere and Health","volume":"19 9","pages":""},"PeriodicalIF":3.1,"publicationDate":"2026-09-14","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148986152","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"环境科学与生态学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Emission characteristics of electrocautery surgical smoke in the operating room: polycyclic aromatic hydrocarbons, metals, persistent free radicals, reactive oxygen species generation, and health risk assessment","authors":"Yu-Chieh Wang,&nbsp;Wei-Cheng Ching,&nbsp;Chia C. Wang,&nbsp;Chon-Lin Lee,&nbsp;Phisit Khemawoot,&nbsp;Wei-Min Ching","doi":"10.1007/s11869-026-02095-1","DOIUrl":"10.1007/s11869-026-02095-1","url":null,"abstract":"<div><p>Surgical smoke generated during operative procedures is a significant source of indoor aerosols in operating rooms, yet its presence is often underestimated, posing occupational exposure risks to healthcare workers. This study reports a systematic characterization of surgical aerosols, with a particular emphasis on fine particles (PM₁). Size-resolved aerosol samples were collected in operating rooms, and the PM₁ fraction was analyzed for carbonaceous composition, polycyclic aromatic hydrocarbons, and heavy metal contents, using electron paramagnetic resonance (EPR) spectroscopy to evaluate the presence of persistent free radicals and the generation of reactive oxygen species (ROS). EPR analysis confirmed the presence of persistent free radicals and the intrinsic ability of surgical aerosols to generate ROS, indicating that surgical smoke should be regarded as a chemically reactive aerosol rather than an inert particulate contaminant. The ROS-generating capacity of PM₁ in the aqueous phase was shown to be 1.03 × 10¹⁸ ± 1.86 × 10¹⁷ spins g⁻¹; this lies between reported values for reference particulate materials such as SRM-2786 urban dust and coal-fired power plant fly ash. Considering the enclosed nature of operating rooms and the repeated exposure of medical personnel to surgical aerosols, sustained inhalation of surgical aerosols may induce severe airway and inflammatory responses, underscoring the requirement for control measures to mitigate occupational exposure in surgical environments. This study demonstrates that surgical procedures generate chemically reactive aerosols in operating rooms with measurable ROS-generating capacity, highlighting that surgical smoke is not merely a mass-based particulate contaminant. Reactive surgical aerosols may act as transient carriers of persistent free radicals and serve as continuous sources of ROS upon interaction with aqueous environments, such as respiratory tract lining fluid. This sustained redox activity underscores the importance of chemical reactivity in determining the health risks of surgical smoke exposure.</p></div>","PeriodicalId":49109,"journal":{"name":"Air Quality Atmosphere and Health","volume":"19 9","pages":""},"PeriodicalIF":3.1,"publicationDate":"2026-09-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148927865","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"环境科学与生态学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Forecasting emergency department visit volume using meteorological variables: a multicenter time-series study","authors":"İbrahim Sarbay,&nbsp;Berhan Coban,&nbsp;Mehmet Göktuğ Efgan,&nbsp;Efe Kanter,&nbsp;Zeynep Özel,&nbsp;Elif Kaymaz,&nbsp;Sevilay Bıyıklı,&nbsp;Mustafa Çalık,&nbsp;Mustafa Agah Tekindal","doi":"10.1007/s11869-026-02097-z","DOIUrl":"10.1007/s11869-026-02097-z","url":null,"abstract":"<div><p>Accurate forecasting of emergency department (ED) volume underpins effective staffing and resource planning. While weather has been proposed as a predictor of demand, its value beyond calendar-based factors remains unclear. This study compared statistical and machine-learning forecasting methods for daily ED attendance and assessed whether meteorological variables add meaningful predictive power. In this multicenter retrospective time-series analysis, daily ED volumes from two tertiary hospitals in İstanbul and İzmir, Türkiye, were examined between October 2023 and June 2025. Meteorological variables included temperature, pressure, humidity, wind speed, and precipitation. Univariate forecasting models were first compared, followed by multivariable analyses using SARIMAX, dynamic regression, random forest, and XGBoost incorporating calendar and weather predictors. Accuracy was assessed via RMSE, MAE, MAPE, sMAPE, and MASE, with Diebold-Mariano tests comparing baseline and weather-augmented models. Robustness was examined with rolling-origin validation, hyperparameter sensitivity, ensemble averaging, and cross-center generalization. After temporal alignment, 625 daily observations per site were analyzed. Among univariate models, NNAR performed best in İstanbul and Theta in İzmir. In the multivariable analyses, calendar-only models gave the best holdout accuracy in both centers (dynamic regression MAPE 5.88% in İstanbul; SARIMAX 4.42% in İzmir). Weather augmentation had inconsistent, model- and center-specific effects: it significantly worsened İstanbul SARIMAX (<i>p</i> = .013) and dynamic regression (<i>p</i> = .015) but significantly improved İstanbul XGBoost (<i>p</i> = .005), with no significant effect in İzmir. Models trained in one center transported poorly to the other, with errors several-fold higher (MAPE up to ~ 39%). Historical attendance and calendar structure were the dominant predictors throughout. Daily ED attendance can be forecast with clinically meaningful accuracy, though optimal models differ by site. Calendar and temporal patterns outweigh meteorological factors in predictive value, and models do not transport across hospitals. Locally tuned, temporally driven forecasting systems may best support ED operational planning. The holdout weather-augmented analyses used observed future meteorological values as exogenous inputs and therefore represent an idealized assessment of the incremental predictive value of weather rather than a fully prospective operational simulation using weather forecasts.</p></div>","PeriodicalId":49109,"journal":{"name":"Air Quality Atmosphere and Health","volume":"19 9","pages":""},"PeriodicalIF":3.1,"publicationDate":"2026-09-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148927526","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"环境科学与生态学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}