Accurately predicting optimal continuous positive airway pressure (CPAP) using traditional PSG parameters remains challenging. Respiratory endotypes derived from PSG can reflect the pathophysiological mechanisms of diseases to a certain extent and are expected to provide a novel perspective for individualized pressure prediction. 114 Chinese adults with obstructive sleep apnea (OSA) underwent two overnight in-laboratory PSG recordings. Optimal CPAP pressure was defined by standardized manual titration performed on the second night. The analysis incorporated three sets of variables: respiratory endotypes, anthropometric indicators, and PSG respiratory parameters. After initial correlation screening, significant variables were entered into a stepwise multiple linear regression model. The dataset was randomly split into training (80%) and validation (20%) sets to assess predictive performance. A total of 114 participants (15 mild, 38 moderate, 61 severe OSA) were included. The prediction model included loop gain (LG1), ventilation capacity (Vpassive, Vmin), BMI, apnea index (AI), mean respiratory event-related oxygen desaturation (RE-OD-Mean), longest apnea duration (Ap-Dur-Long), and respiratory-related arousal index (RR-ArI): CPAP = 0.708 + 1.804×LG1 + 0.023×Vpassive - 0.015×Vmin + 0.117×BMI + 0.003×AI - 0.074×RE-OD-Mean + 0.019×Ap-Dur-Long + 0.043×RR-ArI. The final model demonstrated acceptable predictive performance in the internal validation set (R² = 0.497; RMSE = 1.507 cmH₂O). Among the retained predictors, LG1 showed the largest effect size, with each 1-unit increase associated with an approximately 1.8 cmH₂O higher predicted CPAP pressure. Integrating respiratory endotypes with anthropometric indicators and PSG respiratory parameters provides a mechanism-based framework for CPAP pressure prediction. Collapsibility traits (Vpassive, Vmin) and ventilatory control stability (LG1) substantially enhance predictive accuracy, offering a more individualized approach to PAP therapy.

Influenza A and B viruses are considerable public health threats. Annual seasonal epidemics are driven by antigenic drift and cause an estimated 3-5 million severe cases and 290,000-650,000 deaths annually. The clinical presentation of seasonal influenza is typically an abrupt, uncomplicated acute respiratory illness with full recovery; however, it can lead to severe, life-threatening complications in individuals at high risk. Antigenic shift caused by reassortment of a seasonal influenza A virus with avian and/or swine influenza A viruses has led to unpredictable pandemics. The successful cross-species transmission of influenza A viruses requires key viral adaptations, including changes in receptor specificity and compatibility with host factors such as ANP32A. Host defence involves a layered innate response, which is antagonized by proteins, such as non-structural protein 1, and a robust adaptive immunity comprising cytotoxic CD8⁺ T cells targeting conserved internal proteins and B cells producing neutralizing antibodies. Diagnosis primarily depends on highly sensitive PCR-based methods and multiplexed rapid antigen tests. Prevention involves annually updated seasonal vaccines (including inactivated, live attenuated and protein-based vaccines), while treatment relies on early use of neuraminidase and polymerase acidic protein inhibitors. Research priorities include the development of improved vaccines and a better understanding of influenza virus transmission from animals to humans.

Large language models (LLMs) are increasingly applied in clinical decision support, yet their diagnostic performance in Chinese-language settings and under realistic clinical workflows remains unclear. In particular, how LLMs perform across diseases with different prevalence and under stepwise diagnostic processes has not been well characterized.

This study aimed to evaluate the diagnostic capabilities of LLMs for common diseases and rare diseases using clinical vignettes within a hypothetico-deductive framework and to identify their potential and limitations for clinical diagnosis.

We evaluated 4 Chinese LLMs (Doubao 1.5, DeepSeek-V3, Kimi K1.5, and Leftdoctor GPT 3.5) using 56 clinical cases (28 chronic obstructive pulmonary disease [COPD], and 28 relapsing polychondritis [RP]) sourced from the China Clinical Case Results Database (March 31-April 14, 2025). Patient information was provided incrementally, starting with the initial medical history, followed by physical examination, and laboratory results. Evaluation metrics included top-3 accuracy (RTop3D), top-1 accuracy (RTopD), final diagnostic accuracy (RFA), and mean reciprocal rank (MRR). Statistical analysis was performed using generalized estimating equations (GEE), Friedman tests, and Wilcoxon signed-rank tests with Bonferroni correction. In addition, a qualitative analysis was conducted to characterize recurrent patterns of diagnostic errors.

LLMs demonstrated significantly higher diagnostic accuracy for COPD compared to RP across all metrics (P<.001). Diagnostic accuracy improved after additional clinical information was provided, with the improvement mainly observed in RP cases. In RP, diagnostic accuracy increased from 32.14% to 71.43% for DeepSeek and from 35.71% to 78.57% for Doubao, whereas COPD accuracy remained consistently high across all diagnostic stages (82.14%-92.86%). For COPD, ranking performance was high and comparable among all models (MRR range: 0.82-0.89; P=.71). In RP, diagnostic performance differed significantly among models (MRR range: 0.10-0.39; P<.001). Qualitative analysis showed that COPD errors were mainly related to a failure to recognize specific features, whereas RP errors involved more diverse patterns, particularly the neglect of negative evidence and the failure to recognize specific features.

Chinese LLMs demonstrated relatively strong diagnostic performance for common diseases such as COPD, but lower and less stable performance for rare diseases such as RP. Additional clinical information improved diagnostic accuracy primarily in RP cases, although differences between models remained evident under diagnostically complex conditions. Error patterns in RP cases suggest that current LLMs remain limited in their ability to integrate complex clinical information and exclusionary findings. Careful evaluation and appropriate clinical oversight remain important for their application in clinical practice.

To describe the patterns of medication use among individuals with Covid-19 to treat the disease in Brazil, according to area of residence, demographic and socioeconomic characteristics, religion, and level of trust in scientists.

Epicovid 2.0 National Survey (2024) in 133 municipalities. The analysis included participants who reported having had an episode of Covid-19. The use of antibiotics, azithromycin, antipyretics, chloroquine, and ivermectin was assessed, in addition to the composite outcomes: use of chloroquine and/or ivermectin; chloroquine, ivermectin, and/or azithromycin (CIA); and use of any medication. The frequency of use and 95% confidence intervals were estimated, considering the complex sampling design and sample weighting.

The frequency of use of any medication to treat Covid-19 was 63.0% (95%CI 60.0-65.9). Antipyretics were the most commonly used (53.7%), followed by antibiotics (36.2%) and azithromycin (27.2%). The isolated use of ivermectin was reported by 23.6% and that of chloroquine by 12.1%; the use of ivermectin and/or chloroquine reached 27.8% and that of CIA, 39.0%. Overall, medication use was more frequent in the North, Northeast, and Central-West regions and among women. Use was concentrated among adults aged 40-59, people with low educational attainment, and those with lower incomes. Self-identified black and brown individuals reported higher use of chloroquine. Use of chloroquine, ivermectin, and combinations of these medications was more common among evangelicals (0.05 < p < 0.1) and among people who consulted doctors (p < 0.001). Participants who reported distrusting scientists reported higher use of chloroquine and chloroquine/ivermectin compared to those who reported trusting scientists.

Four out of ten people who reported episodes of Covid-19 consumed medications not recommended by science to treat the disease. Medication use tended to be more frequent among women, in the northern half of Brazil, among evangelicals, and among people with lower income and education levels. These groups should be prioritized in future campaigns on the appropriate use of medications.

A 28-year-old man with HIV (diagnosed eight years prior, last CD4 count 23 cells/mm3) and prior tuberculosis (cured, then confirmedly reinfected with treatment default) presented with cachexia, diarrhea, chest pain, and tachypnea with increased work of breathing that required orotracheal intubation. Admission chest CT showed extensive bilateral consolidations with air bronchograms, suggestive of previous mycobacterial infection and/or possible new bacterial infection. Initially managed as sepsis, blood cultures grew Rhodococcus equi (four samples). Molecular testing confirmed Mycobacterium tuberculosis in sputum and tracheal aspirate, establishing coinfection and exceeding the similarities in acid-fast staining, clinical, and radiological findings between the two organisms. Treatment included antituberculosis therapy associated with clarithromycin and vancomycin (14 days). The patient achieved complete clinical improvement with partial radiological resolution (residual cavitations). He was discharged after 25 days, but was lost to follow-up and died three months later from probable septic shock. This report highlights a rare M. tuberculosis/R. equi coinfection in an immunosuppressed patient, emphasizing diagnostic challenges and the importance of therapeutic adherence in advanced HIV.

to analyze appreciation and coping strategies of children hospitalized for COVID-19 and their primary caregivers during the illness and hospitalization process.

a qualitative and exploratory study, based on the McCubbin & McCubbin model (1993), with 25 participants: ten children hospitalized for COVID-19 and 15 primary caregivers. A semi-structured interview was used, and data were subjected to reflective thematic analysis.

children reported a predominantly negative appreciation of hospitalization, marked by fear, homesickness, pain, and discomfort with hospital routine. Their coping strategies included play, electronic devices, hygiene, and adherence to treatment. Caregivers also expressed negative assessments related to fear of death, isolation, and distress. Their strategies involved spirituality, religiosity, love for their children, and support from professionals and family members.

coping strategies differed between children and caregivers, highlighting the importance of care sensitive to the dyad's needs.

To evaluate the completeness and timeliness of records of severe acute respiratory syndrome (SARS) due to COVID-19 in the SIVEP-Gripe system in Brazil, from 2020 to 2023.

This was a descriptive study based on secondary, public, and anonymized data from SIVEP-Gripe, comprising 2,200,796 SARS due to Covid-19 records. Sixty-six sociodemographic, clinical, hospitalization, outcome, and treatment variables were analyzed. Completeness was classified according to Romero and Cunha (2006). Timeliness was assessed through the intervals between symptom onset and notification, notification and data entry, and notification and case closure, considering historical surveillance parameters (7, 30, and 60 days, respectively).

There was a predominance of variables classified as having very poor completeness throughout the analyzed period, despite a slight improvement in the good and excellent completeness classifications between 2020 and 2023. Regional heterogeneity was observed; Amapá, Mato Grosso do Sul, Santa Catarina, and Paraná achieved higher proportions of satisfactory completion, whereas Tocantins, Pernambuco, Mato Grosso, Rondônia, and Alagoas had the worst performance. Between 2020 and 2023, the average notification time decreased from 13 to 7 days. Data entry was performed on time in over 90% of records in most states, while case closure remained around 86%. Nevertheless, the 90% target for timely notification was not achieved.

Despite improvements in the timeliness of records, completeness remained unsatisfactory and uneven across states. These findings highlight the need to strengthen data entry practices and invest in continuous training and systematic monitoring to ensure more consistent data for surveillance and public health planning.

Early detection of obstructive sleep apnea (OSA) in adults with metabolic dysfunction associated steatotic liver disease (MASLD) remains a clinical challenge. Plasma proteomics provides a non-invasive tool for identifying individuals at elevated risk prior to symptom onset. We quantified 2911 plasma proteins in 17 375 OSA-free MASLD subjects from the UK Biobank (UKB). Participants were split into training and validation sets. Cox regression and Light Gradient Boosting Machine (LightGBM) with forward feature selection were used to identify and rank predictive proteins. Model performance was assessed by area under the receiver operating characteristic curve (AUC) in the validation set, and shapley additive explanations (SHAP) values were used to interpret feature contributions. Among 17 375 MASLD participants, 541 (3.1%) developed subsequent clinical diagnosis of OSA. Seven core proteins (FABP4, PON3, RTN4R, KIAA0319, CFC1, TFRC, CST3) were identified and combined into a protein risk score (ProRS). The ProRS stratified high- and low-risk individuals with C-index 0.727 and 10-year AUC 0.727, outperforming the clinical risk score (C-index 0.548). Integration with clinical factors further improved discrimination (C-index 0.765, 10-year AUC 0.765) and net benefit. The plasma protein-based ProRS predicts future OSA in MASLD and enhances risk stratification beyond clinical variables, supporting its potential for early, non-invasive screening and targeted intervention.

Pseudomonas aeruginosa bacteremic pneumonia carries exceptionally high mortality, yet there is a paucity of genomic characterization of the strains causing this infection. We performed hybrid sequencing (Illumina and Oxford Nanopore) of 12 non-redundant P. aeruginosa isolates from patients with severe bacteremic pneumonia admitted to the intensive care unit of a tertiary hospital between 2015 and 2023. The 12 isolates were assigned to nine distinct sequence types, suggesting that severe bacteremic pneumonia can arise from diverse P. aeruginosa lineages rather than being dominated by a single specialized or high-risk clone. In the combined dataset of our isolates and publicly available Korean P. aeruginosa genomes, type III secretion system exotoxin genotypes exoU and exoS showed a mutually exclusive and phylogenetically segregated distribution, as previously reported, with both genotypes represented among bacteremic pneumonia isolates. Carbapenemase genes were detected in only one isolate, PA22 (ST773), which harboured bla _NDM-1 together with bla _OXA-796 and was the only isolate displaying phenotypic carbapenem resistance and multidrug resistance. To assess the clonal relationship between bla _NDM-1-positive PA22 and the carbapenemase-negative ST773 isolate PA20 and to track the evolution of PA22 resistome within a broader epidemiological context, we investigated the population structure of a global ST773 dataset. Core genome MLST-based minimum spanning trees revealed a deep bifurcation within ST773, separating bla _NDM-1-positive and carbapenemase-negative lineages. Korean ST773 isolates formed two distinct clusters within the NDM-1-positive lineage, with the PA22-containing cluster phylogenetically proximal to isolates from the United States. Within the NDM-1-positive Korean cluster, bla _OXA-796 was located in conserved class 1 integron gene cassette arrays that exhibit ongoing structural diversification among closely related isolates, evidenced by variable integration of IS110 elements. Our findings demonstrate that severe bacteremic pneumonia arises from phylogenetically diverse P. aeruginosa lineages and provide genomic context for the NDM-1-producing ST773 clone that is rapidly emerging in Korea.

Endovascular aneurysm repair (EVAR) is used to treat patients with abdominal aortic aneurysm (AAA) who have suitable anatomy, while complex AAAs often require fenestrated and branched stent grafts, necessitating further refinement of the technique.

To evaluate the safety and efficacy of fenestrated/branched EVAR (F/B-EVAR) assisted by three-dimensional (3D) printing for treating complex AAA.

Retrospective cohort study.

This multicenter retrospective cohort study collected baseline data and clinical outcomes from patients treated with F/B-EVAR between January 2012 and June 2025. The population was divided into the 3D simulation group and the conventional measurement group. The 3D simulation group underwent a simulation to assess fenestration and branch positioning strategies. Study endpoints included procedural success rate, incidence of periprocedural complications and major adverse events, as well as procedural duration and radiation dose.

The mean age of patients was 70.2 ± 8.8 years, with 68.9% being male. Compared to the conventional measurement group, the 3D simulation group had a significantly higher procedural success rate (98.8% vs 91.6%, p < 0.001), shorter procedural duration ((115.2 ± 35.0) min vs (138.7 ± 43.7) min, p < 0.001), and lower radiation dose ((107.2 ± 43.9) mGy vs (144.5 ± 56.3) mGy, p < 0.001). Furthermore, although no significant differences were observed in the incidence of all-cause mortality and major cardiovascular adverse events during the 30-day follow-up between the two groups, the incidence of life-threatening major bleeding, major vascular complications, and acute kidney injury stage ⩾3 was significantly lower in the 3D simulation group.

F/B-EVAR guided by 3D simulation may allow for accurate planning of fenestration and branch positioning, potentially improving procedural success rate and suggesting a potential to enhance the quality of outcomes.