Clinical outcomes of mild-to-moderate mixed aortic valve disease versus isolated aortic stenosis.

Development and International Validation of a Novel Imaging-Based Risk Score (IsCHEMiA) for the Prediction of Poststroke Epilepsy Leung WCY, Tanaka T, Donahue RA, Chan KCW, Wang J, Chung THF, Wong YK, Liu KCR, Leung IYH, Ho RW, Chu FHN, Leung AKL, Teo KC, Pang SYY, Yip EK, Ihara M, Fukuma K, Reeder HT, Chibnik LB, Cole AJ, Lau GKK, Singhal AB. Neurology . 2026;106(2):e214486. Background and Objectives: Stroke is one of the most common causes of adult-onset epilepsy. We aimed to develop a model to predict poststroke epilepsy (PSE) after a first-ever ischemic stroke, incorporating neuroimaging features of incident stroke. Methods: We analyzed clinical and neuroimaging features of patients with first-ever acute ischemic stroke consecutively admitted to Massachusetts General Hospital, United States. We performed competing risk regression with all-cause mortality as a competing event and derived the final multivariable model using backward stepwise elimination by the Akaike Information Criterion. We externally validated the model in three international cohorts in Hong Kong (Queen Mary Hospital [HK-QMH], Ruttonjee Hospital [HK-RH]) and Japan (National Cerebral and Cardiovascular Center) by discrimination and calibration and compared its performance with the SeLECT and SeLECT2.0 scores. Results: We included a final derivative cohort of 1436 patients with a mean age of 67.4 years and a slight male predominance (54.7%), along with a total of 2534 patients in the validation cohorts. PSE, defined as the occurrence or recurrence of unprovoked seizure >7 days after stroke, occurred in 5.5% of the overall study population. Six variables (infarct size [Is], cortical involvement [C], hemorrhagic transformation [H], early seizures [E], MCA involvement [Mi], and age younger than 65 [A]) were independent predictors included in the final model and formed the IsCHEMiA score. Model discrimination was consistent across all cohorts, with c-statistics of 0.870 (United States), 0.852 (HKQMH), 0.857 (HK-RH), and 0.826 (Japan). The model was well calibrated at 1 and 3 years after stroke in the overall validation cohort. The IsCHEMiA score improved the prediction of PSE compared with SeLECT in all cohorts and the overall study population (c-statistic 0.848 vs 0.782, z = 5.170, P < .0001). For example, an IsCHEMiA score of 3 predicts a low risk of PSE at 1 year (2%) and 5 years (6%) while an IsCHEMiA score ≥8 predicts a high risk at 1 year (67%) and 5 years (78%). Discussion: The IsCHEMiA score is an improved and readily applicable predictive model developed and validated using international stroke cohorts in the modern era of reperfusion therapies. It serves as a foundation for personalized management and may guide future clinical trials on antiepileptogenic therapies in acute ischemic stroke.

Periprosthetic joint infection (PJI) is an uncommon but serious complication of total knee arthroplasty (TKA). A previous retrospective cohort study at our institution reported a PJI incidence of 1.34% between 1993 and 2013. The present study aimed to determine whether the incidence of PJI after TKA has changed at our hospital and to evaluate changes in microbiological patterns between 2014 and 2021. In total, 2171 primary TKAs were performed at Queen Mary Hospital in Hong Kong between 1 January 2014 and 31 December 2021. All cases of PJI were identified using the Musculoskeletal Infection Society criteria. Patient demographics, PJI occurrence, and microbiological data were collected and compared with the previously published findings from the 1993-2013 PJI cohort. The incidence of PJI after TKA was 0.64% between 2014 and 2021, representing a significant decrease from the incidence of 1.34% observed at our institution between 1993 and 2013 (P=0.018). There was no significant difference in the incidence of early-onset infection (P=0.095). Methicillin-sensitive Staphylococcus aureus was the most common causative organism, accounting for 57.1% (n=8) of our cohort and 26.5% (n=9) in the previous cohort. The incidence of PJI decreased significantly from 1.34% to 0.64% between the two study periods, suggesting the effectiveness of infection-reduction measures implemented at our institution. Minimal differences were observed in the microbiological patterns of PJI between the cohorts.

Neoadjuvant chemotherapy combined with dual HER2-targeted therapy improves pathological complete response (pCR) rates in early-stage HER2-positive breast cancer. However, data on its efficacy and tolerability in older adults remain limited. To evaluate the efficacy, toxicity, and survival outcomes of neoadjuvant dual anti-HER2 therapy in older adults with HER2-positive early breast cancer. This retrospective cohort included patients with HER2-positive early breast cancer treated with neoadjuvant dual anti-HER2 therapy at Queen Mary Hospital, Hong Kong, between January 2017 and December 2023. Patients were stratified by age (< 65 vs. ≥ 65 years). The primary outcome was pCR. Multivariable logistic regression identified predictors of pCR. Progression-free survival (PFS) was assessed using the Kaplan-Meier method with log-rank testing. Treatment-related toxicities (TRT) were graded according to CTCAE v5.0. A total of 227 patients were included, 43 (18.9%) were aged ≥ 65 years. The pCR rate was 51.2% in older patients and 62.0% in younger patients (P = .194). Higher clinical T-stage was significantly associated with lower pCR (OR = 0.25, 95% CI, 0.06-0.67, P = .02). Older patients experienced more grade ≥ 3 TRT (55.8% vs. 28.3%, P < .001), with neutropenia (32.6% vs. 12.0%) and diarrhoea (16.3% vs. 7.1%) being most common. Unplanned hospitalisations were more frequent (14.0% vs. 4.9%, P = .001). No significant difference in PFS was observed (P = .21). Five-year PFS rates were 87.4% (95% CI, 76.3%-100%) in older patients and 95.2% (95% CI, 92.0-98.6%) in younger patients. Older patients achieved pCR and PFS comparable to younger counterparts but experienced higher toxicity and hospitalisation rates. Age-specific and de-escalation strategies warrant further study.

BackgroundThe role of the Advanced Lung Cancer Inflammation Index (ALI) in chronic obstructive pulmonary disease (COPD) remains unclear, although it has been utilized to investigate various non-malignant conditions.MethodsA prospective study involving Chinese patients with COPD was carried out in Hong Kong to examine the relationship between baseline ALI levels and the risk of acute exacerbations of COPD (AECOPD). ALI was evaluated across quartiles. Patients were prospectively recruited from respiratory clinic in Queen Mary Hospital and Grantham Hospital in 2021, follow up with patients was done until 8th March 2025 or the death date, whichever is earlier.ResultsAmong 272 Chinese COPD patients recruited, 138 of them had moderate to severe AECOPD and 66 patients died in the follow-up period. Those in the Q1 ALI, when compared with Q4 (highest quartile), had significantly shorter time to moderate to severe AECOPD with adjusted hazard ratio of (aHR) 2.17 (95% CI = 1.29–3.65, p = 0.011), severe AECOPD (aHR 2.05, 95% CI = 1.18–3.55, p = 0.011) and overall survival (aHR 2.73, 95% CI = 1.21–6.15, p = 0.015). The same phenomenon was also observed in the patient subgroup with baseline blood eosinophil counts <300 cells/μL.ConclusionIn this prospective study, it suggested that ALI can serve as a biomarker to predict the risk of moderate to severe AECOPD, as well as severe AECOPD and mortality. The phenomenon was also observed in the non-eosinophilic subgroup. This can allow clinicians to use this simple and repeatable biomarker as a way to prognosticate COPD patients and estimate AECOPD risks.

Stroke is one of the most common causes of adult-onset epilepsy. We aimed to develop a model to predict poststroke epilepsy (PSE) after a first-ever ischemic stroke, incorporating neuroimaging features of incident stroke. We analyzed clinical and neuroimaging features of patients with first-ever acute ischemic stroke consecutively admitted to Massachusetts General Hospital, United States. We performed competing risk regression with all-cause mortality as a competing event and derived the final multivariable model using backward stepwise elimination by the Akaike Information Criterion. We externally validated the model in 3 international cohorts in Hong Kong (Queen Mary Hospital [HK-QMH], Ruttonjee Hospital [HK-RH]) and Japan (National Cerebral and Cardiovascular Center) by discrimination and calibration and compared its performance with the SeLECT and SeLECT2.0 scores. We included a final derivative cohort of 1,436 patients with a mean age of 67.4 years and a slight male predominance (54.7%), along with a total of 2,534 patients in the validation cohorts. PSE, defined as the occurrence or recurrence of unprovoked seizure >7 days after stroke, occurred in 5.5% of the overall study population. Six variables (infarct size [Is], cortical involvement [C], hemorrhagic transformation [H], early seizures [E], MCA involvement [Mi], and age younger than 65 [A]) were independent predictors included in the final model and formed the IsCHEMiA score. Model discrimination was consistent across all cohorts, with c-statistics of 0.870 (United States), 0.852 (HK-QMH), 0.857 (HK-RH), and 0.826 (Japan). The model was well calibrated at 1 and 3 years after stroke in the overall validation cohort. The IsCHEMiA score improved the prediction of PSE compared with SeLECT in all cohorts and the overall study population (c-statistic 0.848 vs 0.782, z = 5.170, p < 0.0001). For example, an IsCHEMiA score of 3 predicts a low risk of PSE at 1 year (2%) and 5 years (6%) while an IsCHEMiA score ≥8 predicts a high risk at 1 year (67%) and 5 years (78%). The IsCHEMiA score is an improved and readily applicable predictive model developed and validated using international stroke cohorts in the modern era of reperfusion therapies. It serves as a foundation for personalized management and may guide future clinical trials on antiepileptogenic therapies in acute ischemic stroke.

Single-center retrospective study. By utilizing three-dimensional (3D) reconstruction models, our study aimed to investigate the three-dimensional changes in vertebral body tethering (VBT) and assess the relationship between axial-plane parameters and postoperative outcomes. Previous studies mainly focused on coronal plane correction but lacked investigation on axial plane changes following VBT. We included consecutive patients who underwent VBT in our institution (Queen Mary Hospital and Duchess of Kent Children's Hospital, Hong Kong) from February 2019 to April 2024. We used EOS radiographs to generate 3D reconstruction models, and parameters were analyzed preoperatively, immediately postoperatively, 1 year postoperatively, and 2 years postoperatively. The primary outcomes were changes in coronal, axial, and sagittal profiles at different time points. Secondary outcomes included the relationship between axial parameters with short-to-medium term changes in coronal/sagittal profiles. We included 44 patients (seven males, 37 females) with 58 instrumented curves, with an average follow-up of 36.3±17.1 months. The mean Cobb angle, apical vertebral rotation (AVR), and maximal vertebral rotation (MVR) improved from 48.0°±10.7°, 9.1°±5.7°, and 13.4°±5.7° preoperatively to 22.3°±8.9°, 6.2°±4.8°, and 9.8°±4.3° postoperatively, respectively, with correction maintained at 2 years. Preoperative AVR, MVR, and intraoperative derotation were significantly correlated with 1-year and 2-year correction rate and curve regression (Pearson correlation coefficient [r ]=0.35-0.63; p <0.001). Multivariate analysis confirmed AVR derotation and preoperative MVR as significant predictors for the 1-year correction rate. Tether breakage occurred in 27.6% (16/58) of patients. VBT was effective in correcting coronal and axial deformity at 2 years, but most correction occurred intraoperatively. Axial parameters were predictive for postoperative outcomes, with increased preoperative rotation associated with greater coronal correction. More aggressive derotation corresponded to greater correction. To improve surgical outcomes, clinicians should aim to achieve adequate correction by screw positioning and appropriate tensioning.

510 Background: Radiological progression patterns have long been recognized as prognostic indicators in hepatocellular carcinoma (HCC) treated with tyrosine kinase inhibitors. However, their impact in the era of immune checkpoint inhibitors (ICIs) remains underexplored. This study investigates the prognostic significance of progression patterns in advanced HCC patients receiving ICI-based therapies. Methods: We conducted a retrospective cohort study of 293 patients with radiologically confirmed disease progression following ICI treatment at Queen Mary Hospital between January 2015 and September 2023. Progression patterns were categorized as intrahepatic growth, new intrahepatic lesions, extrahepatic growth, and new extrahepatic lesions. Multivariate Cox regression analyses were performed to assess their impact on overall survival (OS) and post-progression survival (PPS), adjusting for ALBI score and AFP levels. Results: Intrahepatic growth was significantly associated with poorer OS in both the first-line (HR 1.905, 95% CI 1.268–2.861; p = 0.002) and overall cohorts (HR 1.632, 95% CI 1.203–2.213; p = 0.002). Extrahepatic growth consistently predicted reduced OS (HR 1.728, 95% CI 1.305–2.290; p &lt; 0.001) and PPS (HR 1.771, 95% CI 1.330–2.360; p &lt; 0.001) across all treatment subgroups. New extrahepatic lesions were associated with inferior OS in the overall (HR 1.402, 95% CI 1.054–1.865; p = 0.020) and first-line populations (HR 1.556, 95% CI 1.060–2.285; p = 0.024), and with poorer PPS in patients receiving ICI combined with anti-VEGF agents (HR 1.842, 95% CI 1.089–3.116; p = 0.023). In contrast, new intrahepatic lesions did not demonstrate significant prognostic impact on PPS in adjusted analyses (HR 0.938, 95% CI 0.708–1.243; p = 0.658). Conclusions: In advanced hepatocellular carcinoma treated with immune checkpoint inhibitors, intrahepatic growth and extrahepatic expansion—particularly of pre-existing lesions—are independently associated with poorer overall and post-progression survival. These findings highlight the prognostic relevance of radiological progression patterns in the immunotherapy era and suggest that both intra- and extrahepatic disease dynamics warrant close surveillance. Integrating systemic therapies with timely locoregional interventions may offer a path to improved outcomes in this patient population.

Hand hygiene is vital for infection prevention, yet compliance among medical staff remains low. This study assessed whether training medical students using a multimodal educational intervention, including their roles as 'hand hygiene covert observers' (HHCOs) in their pre-internship phase, could improve hand hygiene compliance during their internship. A retrospective study was conducted at Queen Mary Hospital, Hong Kong, comparing two consecutive cohorts of medical interns: cohort A (historical control, starting July 1, 2024) and cohort B (intervention group, starting July 1, 2025). Cohort B participated in a half-day training workshop and covertly observed hand hygiene compliance as HHCOs during a 19-day pre-internship clinical attachment in June 2025. Hand hygiene compliance observed by HHCOs was compared with infection control nurses (ICNs) observations collected concurrently. Additionally, hand hygiene compliance among cohorts A and B was monitored by ICNs using World Health Organization audit tools. Baseline knowledge and attitudes of cohort B regarding hand hygiene were assessed by questionnaire. Seventy-four pre-interns in cohort B completed the baseline questionnaire; 38 (51.4%) identified alcohol-based hand rub (ABHR) as the most effective hand hygiene method, while 36 (48.6%) selected soap and water. Positive attitudes were evident, with 54 (73.0%) strongly agreeing on hand hygiene's role in preventing healthcare-associated infections and 52 (70.3%) strongly agreeing that compliance impacts patient safety. During the pre-internship clinical attachment, hand hygiene compliance observed by HHCOs among cohort A was significantly higher than that observed by ICNs (96%, 682/713 vs. 59%, 144/244; p < 0.001). Comparison of ICN-observed compliance showed a non-significant increase for cohort B versus cohort A (66%, 230/348 vs. 58%, 156/267; p = 0.051). However, cohort B demonstrated a significantly higher proportion of hand hygiene episodes using ABHR compared to cohort A (90%, 208/230 vs. 79%, 123/156; p = 0.001). Engaging pre-interns in a multimodal educational intervention, including their roles as covert observers, did not significantly increase overall hand hygiene compliance compared to historical controls. However, there was a notable rise in ABHR use among the intervention group. This approach may promote awareness and foster a culture of patient safety.

Clinical and molecular epidemiology of Streptococcus pneumoniae infections during and in early recovery phase of COVID-19 pandemic in Hong Kong.

Obesity is a growing problem worldwide. It is well reported that obesity is associated with various degrees of respiratory function compromise, which could be contributed by mechanical factors as well as co-morbid diseases such as asthma. While spirometry is the gold standard for lung function measurement, it does have the drawback of only being able to measure the global function of the lungs and is uncomfortable for those with difficulty using a large degree of effort. A prospective cross-sectional study was conducted in Queen Mary Hospital, Hong Kong to recruit Chinese obese and non-obese subjects, to compare and correlate the lung function parameters measured by electrical impedance tomography (EIT) and spirometry. Obesity was diagnosed based on the local definition of body mass index of 25 kg/m2 or above, using the cut-off value for Asian population. A total of 42 obese subjects and 35 non-obese subjects were included in this study. The spirometric values (Litres) measured by spirometry and EIT demonstrated good correlation among obese and non-obese subjects, with Pearson correlation coefficient of 0.705 and 0.673 for FEV1 among obese and non-obese subjects, p-value < 0.01. The Pearson correlation coefficients were 0.568 and 0.541 for FVC among obese and non-obese subjects respectively, p-value < 0.01. EIT technology has comparable performance in lung function measurement among obese and non-obese subjects. It can suggest that EIT is a potential tool for lung function measurement in obese subjects as in non-obese subjects.Supplementary InformationThe online version contains supplementary material available at 10.1038/s41598-025-29559-1.

Lupus low disease activity state (LLDAS) is a validated treatment target in systemic lupus erythematosus (SLE), but limited studies have explored the role of LLDAS in lupus nephritis (LN). This study aims to investigate the frequency and predictors of LLDAS attainment and its benefit on LN relapse and renal function preservation in patients with LN. Patients with LN during 2010 to 2020 in Queen Mary Hospital and Pamela Youde Nethersole Eastern Hospital were included in the discovery cohort and validation cohort, respectively. Complete renal response (CRR), partial renal response (PRR), LLDAS, and Definition Of Remission In SLE (DORIS) remission were assessed at 12 months. Regression analysis was performed to identify risk factors of LN relapse. Receiver operating characteristic (ROC) curves were used to evaluate target attainment and long-term kidney function. A total of 245 patients with LN (discovery cohort n = 143 and validation cohort n = 102) were included. At 12 months, 57 of 143 (40%), 14 of 143 (10%), 70 of 143 (49%), and 15 of 143 (10%) patients achieved CRR, PRR, LLDAS, and DORIS remission, respectively. Attainment of both CRR/PRR and LLDAS at 12 months was associated with best relapse-free survival (P < 0.001). Multivariate analysis showed independent association of CRR/PRR and LLDAS with LN relapse risk reduction (CRR/PRR: hazard ratio [HR] 0.31, P = 0.007; LLDAS: HR 0.38, P = 0.029). LLDAS attainment predicts renal function preservation with satisfactory performance in both the discovery and validation cohorts (area under the curve of the ROC 0.71). LLDAS is an attainable target in LN comparable to CRR/PRR. Attainment of both targets is associated with additional benefits on relapse risk reduction. Early LLDAS attainment is associated with renal function preservation.

Non‐small cell lung cancer (NSCLC) is the leading cause of brain metastases (BMs) and is characterized by a poor prognosis and limited response to standard treatments. Multi‐omics sequencings, integrating spatial transcriptomics, metabolomics, single‐cell RNA sequencing, bulk proteomics, and metabolomics, are conducted to analyze tumor and blood specimens from 34 patients with NSCLC with or without BMs from the Xiangya Hospital NSCLC (XY‐NSCLC) and Queen Mary Hospital NSCLC (QMH‐NSCLC) cohorts. This investigation identified LOX+ Malig‐5 cells as metastasis‐initiating cells (MICs) that are significantly associated with poor prognosis. MICs colocalize with specific neutrophil subtypes, which facilitate the formation of neutrophil extracellular traps (NETs) within the metastatic niche. Mechanistically, a NET‐KRT10 signaling axis that mediates the interaction between NET‐releasing neutrophils and LOX+ Malig‐5 cells is discovered, thereby promoting epithelial–mesenchymal transition (EMT) and metastasis. Furthermore, metabolic profiling reveals elevated palmitic acid levels in the resulting metastatic niche, which emerges as a crucial metabolic driver in BMs. Using an AI‐driven prediction model and in vitro/in vivo assays, fatty acid synthase inhibitor TVB‐2640 is identified as a potential therapeutic agent for disrupting metabolic vulnerability and suppressing NSCLC BMs. These findings provide novel insights into NET‐dependent cellular interactions that sustain the pro‐metastatic microenvironment underlying NSCLC BMs, offering robust development of novel metabolism‐based therapeutic strategies to combat this lethal complication.

Abstract Brain metastases are a common complication in cancer patients. Stereotactic radiosurgery (SRS)/ hypofractionated stereotactic radiotherapy (SRT) are more preferred over whole brain radiotherapy (WBRT) in managing this group of patients following surgery. This single center study included 70 patients with 72 resected brain metastases treated with postoperative SRS/ SRT between July 2017 and May 2024 in Queen Mary Hospital, Hong Kong. Study endpoints included in-field and out-field intracranial progression-free survival (PFS), overall survival (OS) and incidence of radionecrosis. Prognostic factors were analyzed using Cox regression. Recursive partitioning Analysis (RPA) was performed to develop an OS prognostic model. The median follow-up was 15.1 months. 1-year and 2-year in-field PFS was 96.2% and 90.2% respectively. 1-year and 2-year out-field PFS was 69.3% and 64.4% respectively. Prior cranial radiotherapy was associated with worse out-field PFS (HR 4.23, 95% CI 1.24-14.41, p=0.02). OS rates were 67.5% at 1 year, 46.7% at 2 years. Performance status, time from surgery to radiotherapy of ≤56 days and biologically effective dose using α/ß of 10 (BED10) ≥51Gy were significant predictors of OS. Radiation necrosis occurred in 7 patients (10%) with median onset was 25.0 months. Our RPA model stratified patients into 3 prognostic groups: RPA-I (ECOG &amp;lt;2 &amp; BED10 ≥51Gy; median OS 43.4 months, 95% CI 22.3 months - Not reached), RPA-II (ECOG &amp;lt;2 &amp; BED10&amp;lt;51Gy; median OS 11.9 months, 95% CI 3 months - 19.2 months) and RPA-III (ECOG ≥2 regardless of BED10 delivered; median OS 4.1 months, 95% CI 1.2 months – 22.0 months) (p&amp;lt;0.001). RPA-II and RPA-III had increased risk of death compared with RPA-I. This model outperformed existing models after bootstrapping validation. Postoperative SRS/ SRT is an effective treatment for brain metastases. Appropriate patient selection, timely initiation and sufficient dose improves outcomes. Our RPA system demonstrates robust result for prognostication.

Asciminib as monotherapy or adjunctive therapy for chronic myeloid leukaemia in blast phase

BackgroundUndermining Iatrogenic Coronary Obstruction With Radiofrequency Needle (UNICORN) was developed to mitigate coronary obstruction risk during valve-in-valve transcatheter aortic valve replacement (ViV-TAVR) with balloon-expandable valves (BEVs). A modified UNICORN with recrossing technique was subsequently described for self-expanding valves (SEVs).ObjectivesThis study aimed to assess early outcomes of the UNICORN technique.MethodsThe UNICORN Hong Kong Registry was a single-center, retrospective observational study at Queen Mary Hospital, Hong Kong, including consecutive patients undergoing ViV-TAVR between July 2024 and August 2025. Patients were treated with either UNICORN using BEV or UNICORN with recrossing using SEV. The primary endpoint was procedural success, defined as successful leaflet traversal, intended leaflet laceration, and transcatheter heart valve implantation without procedural mortality, coronary obstruction, or emergency surgery. Secondary outcomes included 30-day safety events per Valve Academic Research Consortium-3.ResultsSeventeen patients (41.2% male; 78.3 ± 6.61 years) underwent ViV-TAVR: 8 with BEV and 9 with SEV using the recrossing technique. All SEV patients had prior surgical valves, whereas in the BEV group 50.0% had surgical valves and 50.0% had transcatheter valves. Coronary ostium at risk was left in 58.8%, right in 29.4%, and bilateral in 11.8%. Median coronary height was 7.80 [7.00-8.50] mm and mean virtual transcatheter heart valve to coronary distance was 4.56 ± 1.48 mm. One patient required bileaflet modification with concomitant UNICORN and Bioprosthetic or Native Aortic Scallop Intentional Laceration to Prevent Iatrogenic Coronary Artery Obstruction. Procedural success was achieved in all patients with no coronary obstruction. At 30 days, 1 patient (5.88%) required pacemaker implantation.ConclusionsThe UNICORN technique was feasible for preventing coronary obstruction during ViV-TAVR and warrants further evaluation in larger trials.

BackgroundThe role of technology in disease management has gained significant attention, mobile health technology for obstructive sleep apnea (OSA) remains underdeveloped, with limited research exploring its applications. This study explored OSA individuals’ health literacy, perceived needs for improvement, information preferences, and acceptance of smartphone-based messaging interventions.MethodsThe study comprised two parts. Part 1 was a cross-sectional, self-administered online survey using non-probability convenience sampling conducted in the inpatient ward and outpatient clinic of Queen Mary Hospital, Hong Kong, from May to July 2020. The questionnaire assessed health literacy, perceived needs, and views on smartphone-based messaging interventions to enhance self-management of OSA. Written consent was obtained, and the survey took approximately 10 minutes to complete. Part 2 involved a focus group interview in August 2020 with nine randomly selected participants from Part 1, exploring their views on perceived needs and smartphone-based messaging interventions.ResultsA total of 108 participants (82% men; 40.7% aged ≥60 years) were included. Participants self-reported their OSA severity as mild (21.3%), moderate (17.6%), and severe (23.1%), while 38% were unaware of their severity. The results showed low OSA-related health literacy among participants, with no significant differences by age or years since diagnosis. Most participants (76.8%) expressed a strong desire to improve their health. High demand for OSA-related information was reported, including sleep quality (83.3%), OSA-related knowledge (82.4%), weight control (81.5%), habit improvement (78.7%), and emotional management (66.7%). The messaging intervention was well accepted, with high feasibility (82.4%) and usability (72.2%).ConclusionsThis study highlights low OSA-related health literacy and a strong demand for authoritative health information. Most found that the messaging intervention was feasible and useful in improving health-related habits and OSA management. The study underscores the need for tailored health education, personalized interventions targeting perceived needs, and innovative messaging solutions to enhance treatment adherence and self-management in OSA individuals.

Background: Infection prevention and control (IPC) in hospitals relies heavily on infection control nurses (ICNs) who manage complex consultations to prevent and control infections. This study evaluated large language models (LLMs) as artificial intelligence (AI) tools to support ICNs in IPC decision-making processes. Our goal is to enhance the efficiency of IPC practices while maintaining the highest standards of safety and accuracy. Methods: A cross-sectional benchmarking study at Queen Mary Hospital, Hong Kong assessed three LLMs—GPT-4.1, DeepSeek V3, and Gemini 2.5 Pro Exp—using 30 clinical infection control scenarios. Each model generated clarifying questions to understand the scenarios before providing IPC recommendations through two prompting methods: an open-ended inquiry and a structured template. Sixteen experts, including senior and junior ICNs and physicians, rated these responses on coherence, conciseness, usefulness and relevance, evidence quality, and actionability (1–10 scale). Quantitative and qualitative analyses assessed AI performance, reliability, and clinical applicability. Results: GPT-4.1 and DeepSeek V3 scored significantly higher on the composite quality scale, with adjusted means (95% CI) of 36.77 (33.98–39.57) and 36.25 (33.45–39.04), respectively, compared with Gemini 2.5 Pro Exp at 33.19 (30.39–35.99) (p < 0.001). GPT-4.1 led in evidence quality, usefulness, and relevance. Gemini 2.5 Pro Exp failed to generate responses in 50% of scenarios under structured prompt conditions. Structured prompting yielded significant improvements, primarily by enhancing evidence quality (p < 0.001). Evaluator background influenced scoring, with doctors rating outputs higher than nurses (38.83 vs. 32.06, p < 0.001). However, a qualitative review revealed critical deficiencies across all models, for example, tuberculosis treatment solely based on a positive acid-fast bacilli (AFB) smear without considering nontuberculous mycobacteria in DeepSeek V3 and providing an impractical and noncommittal response regarding the de-escalation of precautions for Candida auris in Gemini 2.5 Pro Exp. These errors highlight potential safety risks and limited real-world applicability, despite generally positive scores. Conclusions: While GPT-4.1 and DeepSeek V3 deliver useful IPC advice, they are not yet reliable for autonomous use. Critical errors in clinical judgment and practical applicability highlight that LLMs cannot replace the expertise of ICNs. These technologies should serve as adjunct tools to support, rather than automate, clinical decision-making.

Asthma is associated with both airway and systemic inflammation as well as non-respiratory adverse outcomes. However, data regarding its impact on long-term renal outcomes is lacking. We classified all asthma patients who were followed at Queen Mary Hospital in 2017 into eosinophilic or non-eosinophilic phenotypes based on their highest blood eosinophil counts (BEC) during stable state in the year (≥ 300 or < 300 cells/mm3 respectively) and prospectively evaluated their clinical outcomes in the subsequent 5 years. The relationship between patient phenotypes and the long-term renal outcomes were assessed. Five hundred and four asthma patients with baseline Stage 1 to 3 chronic kidney disease were included [296 (58.7%) and 208 (41.3%) in eosinophilic and non-eosinophilic groups respectively]. Among patients with baseline renal function at CKD stage 1 to 3, one hundred and four patients (20.6%) had renal progression in this cohort (56 patients (26.9%) vs. 48 patients (16.2%) in the non-eosinophilic and eosinophilic groups respectively). Patients with non-eosinophilic asthma showed increased risks of renal progression over 5 years of follow-up [adjusted odds ratio (aOR) 2.615, 95% CI 1.151–5.942 p = 0.022] and more rapid eGFR decline (−4.29 ± 3.48 mL/min/1.73m2/year vs. −3.48 ± 3.07 mL/min/1.73m2/year, p = 0.007) than those with eosinophilic phenotype. Patients who developed renal progression had higher risk of death [adjusted hazard ratio (aHR) 1.614 (95% CI 1.041–2.502); p = 0.032]. Progressive renal function deterioration is prevalent amongst asthma patients, and those with non-eosinophilic phenotype are at risk of renal progression.Supplementary InformationThe online version contains supplementary material available at 10.1038/s41598-025-18035-5.

Fractional exhaled nitric oxide (FENO) serves as a marker of eosinophil-mediated airway inflammation and has been used in asthma diagnosis, phenotyping, and guidance regarding selection and adjustment of asthma therapy. Studies suggested that FENO correlated with asthma symptoms, peripheral blood eosinophil level, blood IgE level, and spirometry indicators of airway obstruction. However, the results are inconsistent across studies. A prospective cross-sectional study was conducted in Queen Mary Hospital among adult patients with severe asthma. Patients had spirometry with bronchodilator reversibility and same-day FENO. Asthma control test (ACT) score and blood eosinophil and total IgE levels were measured within 4 weeks of FENO and spirometry. The primary outcome was the correlation of FENO and spirometric values. The secondary outcomes included the correlation of FENO with ACT score, blood eosinophil, and total IgE levels. One hundred thirty-five severe asthma patients with FENO performed were included in the study. FENO was negatively correlated with pre-bronchodilator FEV1 (L) (r = -0.188, p = 0.029), pre-bronchodilator FEV1 (% predicted) (r = -0.169, p = 0.050), pre-bronchodilator FEV1/FVC ratio (r = -0.269, p = 0.002), and post-bronchodilator FEV1/FVC (r = -0.215, p = 0.018). FENO was positively correlated with bronchodilator reversibility (mL) (r = 0.248, p = 0.006) and bronchodilator reversibility (%) (r = 0.823, p = 0.002), baseline blood eosinophil level by absolute cell count (r = 0.308, p < 0.001) and by percentage (r = 0.361, p < 0.001). In adult patients with severe asthma, FENO might have a negative correlation with the FEV1, FEV1/FVC ratio, and a positive correlation with bronchodilator reversibility, as well as with blood eosinophil levels.