Background Bloodstream infection (BSI) contributed significant mortality among patients in the intensive care unit (ICU). Traditional machine learning (ML) models often struggle to effectively capture complex temporal dependencies in high-dimensional data. The aim of this study was to develop a deep learning model transformer in the prediction of ICU-acquired BSI based on time series data. Methods Patients' electronic health records, whose all blood cultures (BC) collected 48 h after admission to the ICU, were extracted from Medical Information Mart for Intensive Care IV (MIMIC IV). The synthetic minority over-sampling technique (SMOTE) was applied to balance the dataset. We collected age, gender, vital signs and laboratory measures for consecutive 24 h with 1 hour interval. We also set three prediction windows (0, 12 and 24 h) to investigate the ability of early detection of the ML. The performances of the transformer and the CatBoost were evaluated by discrimination and calibration. Shapley Additive exPlanation (SHAP) was employed to identify key features. Results A total of 2408 patients were included in the study, of which 149 (6.2%) had an ICU-acquired BSI. The transformer model outperformed CatBoost at all prediction windows. At the 24-hour window, the Transformer achieved an AUROC of 0.918 and an AUPRC of 0.915, while CatBoost performance declined significantly with earlier prediction. SHAP values suggested that glucose, bicarbonate, mean blood pressure, temperature and blood urea nitrogen were top five early predictors. Conclusion The deep learning transformer using time series data demonstrates strong potential as a clinical decision support tool.

Tracheostomy is an important therapeutic measure to relieve airway obstruction and ensure effective ventilation in patients. However, the procedure bypasses the natural upper airway, compromising its normal defensive functions, which can lead to drying of secretions, crust formation, potential tube occlusion, and ultimately pulmonary infection. Adequate airway humidification can help dilute sputum, facilitate sputum drainage, and ultimately reduce the incidence of airway-related complications. Although several airway humidification methods have been reported, their outcomes remain unsatisfactory. To address this, the intensive care unit staff of Hangzhou First People's Hospital, designed a multifunctional continuous humidification device consisting of a cover, connecting tube, and nebulizer for airway humidification, which has been granted a national utility model patent of China (patent number: ZL 2021 2 0079525.4). A total of 48 patients requiring oxygen therapy via a tracheostomy at the intensive care unit of Hangzhou First People's Hospital between August 2023 and June 2024 were selected as the study subjects. Using a random number table, they were assigned in a 1 : 1 ratio to an experimental group (n=24) and a control group (n=24). From the initiation of the transition from mechanical ventilation weaning to continuous oxygen inhalation via the tracheostomy, the experimental group received humidification using the multifunctional continuous humidification device, while the control group received humidification using a traditional humidifier. The daily humidification protocol was consistent for both groups and was implemented by nursing staff who had undergone homogenized training. General patient data, the incidence of complications during humidified oxygen therapy via tracheostomy, nursing staff satisfaction, and sputum viscosity grading on day 1, 7, and 14 of airway humidification were compared between the two groups. The results showed no statistically significant differences in gender, age, disease types, or pre-humidification sputum viscosity grades between the two groups. The incidence of complications in the experimental group was lower than that in the control group during the treatment [crust formation: 4% (1/24) vs. 54% (13/24); airway mucosal injury: 8% (2/24) vs. 42% (10/24); pulmonary infection: 4% (1/24) vs. 54% (13/24)]. Nursing staff satisfaction was also found to be significantly higher in the experimental group [100% (24/24) vs. 17% (4/24)]. All these differences were statistically significant (all P<0.05). The sputum viscosity grades on day 1, 7, and 14 were also significantly more favorable in the experimental group than in the control group (all P<0.05). These findings indicate that the application of this device in tracheostomy patients effectively maintained airway humidification, ensured respiratory tract patency, improved airway clearance efficacy, reduced the frequency of sputum suction, and prevented airway complications, and enhances nursing efficiency, demonstrating good potential for clinical promotion.

Objective:Diabetic nephropathy (DN) occurs in nearly 40% of Type 2 diabetes mellitus (T2DM) patients, and there is a positive correlation between DN and terminal renal disease. Thus, exploring novel biomarkers is vital to facilitate early diagnosis and intervention in DN patients; however, indicators of DN are still scant. Since the microRNA-126 (miR-126) may be related to the occurrence of diabetes, we aim to assess the association between miR-126 and DN. Methods:This is a prospective cohort design and a nested case-control approach study that enrolled 300 individuals (100 T2DM patients, 100 DN patients, and 100 controls). miR-126 expression was analyzed by quantitative real-time PCR (qPCR) and compared among three groups. The overall survival (OS) was presented by Kaplan-Meier analysis. The area under the curve (AUC) was used to evaluate the potential of miR-126 as a biomarker for DN. Result:DN patients, compared with T2DM and controls, had lower miR-126 content (p < 0.001), and miR-126 levels significantly decreased following a decreasing estimated glomerular filtration rate (eGFR) (r = 0.65, p < 0.001). Moreover, significant differences were also found in OS among quartiles of serum miR-126 level (p for trend < 0.001). In addition, the AUC for diagnosis DN from T2DM patients was found to be 0.804 (95% CI, 0.745-0.863), with a sensitivity of 83.0% and a specificity of 63.0%. Conclusion:This study provides evidence to clarify that decreased levels of miR-126 were linked to an increased susceptibility to developing DN compared with healthy volunteers. More importantly, the diagnostic role of miR-126 remained significant in differentiating DN from T2DM patients.

To explore the associations between temperature trajectories and in-hospital mortality and renal replacement therapy in patients with sepsis-associated acute kidney injury (SA-AKI). By using data from the Medical Information Mart for Intensive Care (MIMIC)-IV, participants were divided into three groups (≤ 36 °C, 36-38 °C, ≥ 38 °C). We identified body temperature trajectories by a latent class mixed model and explored the associations of these trajectories with in-hospital mortality using Cox hazard proportional regression models, further exploring the associations with renal replacement therapy using logistic regression models. Total 1,831 in-hospital deaths during 9,760 person-years of follow-up were documented. In the hypothermia group, five different temperature trajectory classes were identified: L1, L2, L3, L4, and L5. Similarly, four trajectory classes (M1, M2, M3, and M4) emerged in the normal temperature group, whereas the hyperthermia group presented four distinct trajectory classes (H1, H2, H3, and H4). Compared with patients with the M3 trajectory, those with the L1 (hazard ratio [HR]: 2.41, 95% confidence interval [CI]: 1.58-3.66), L2 (HR: 1.48, 95% CI 1.11-1.97), L3 (HR: 1.27, 95% CI 1.01-1.59), L4 (HR: 1.29, 95% CI 1.08-1.54), and M1 (HR: 1.29, 95% CI 1.06-1.57) trajectories were at greater risk of in-hospital mortality. For patients with different baseline temperatures, the L1 (HR: 1.95, 95% CI 1.19-3.18), M1 (HR: 1.28, 95% CI 1.05-1.56), and H4 (HR: 2.37, 95% CI 1.05-5.36) trajectories were related to an elevated risk of in-hospital mortality. The study suggests that early body temperature trajectories are linked to increased in-hospital mortality risk in patients with SA-AKI.

Sepsis-associated acute kidney injury (S-AKI) is a grave clinical condition with a poor prognosis. Prior studies have observed a potential correlation between copper (Cu) and S-AKI, though the underlying mechanisms remain to be fully elucidated. The present study aims to investigate the underlying mechanism between Cu and renal dysfunction under sepsis conditions. Utilizing a cecal ligation and puncture (CLP) murine model, we observed a substantial upregulation of inflammatory factors and deteriorated renal function (p<0.05), concomitant with an augmented mortality rate (p<0..05). Concurrently, the renal tissue exhibited a series of pathological manifestations, including fibrosis and cell death. Concurrently, the expression of the cuproptosis-regulating protein FDX-1 (p<0.05) was found to be elevated in CLP mice. Collectively, these data indicated the activation of cuproptosis in CLP-induced S-AKI. Intriguingly, the application of abscisic acid (ABA), a FDX-1 inhibitor, led to a substantial alleviation of mortality rate and inflammatory factor level (p<0.05). Furthermore, a significant alleviation was observed in renal pathology and biofunction (p<0.05). Collectively, these observations suggest that ABA exerts a protective effect on renal function during cuproptosis promoted S-AKI. Consequently, it can be concluded that ABA has the potential to reverse cuproptosis in S-AKI by inhibiting FDX-1. ### Competing Interest Statement The authors have declared no competing interest.

Extracorporeal blood purification is a technique for reducing elevated levels of inflammatory mediators and/or endotoxins in critically ill patients with sepsis or other hyperinflammatory conditions. The oXiris filter combines endotoxin adsorption, cytokine adsorption, hemofiltration, and anti-thrombosis, and an emerging body of evidence demonstrates its use in critical care patients with hyperinflammatory conditions and acute kidney injury (AKI). A group of Asia-Pacific experts convened to formulate consensus statements for the use of the oXiris filter based on a comprehensive review of publications. After evaluation using a Delphi methodology, 17 statements achieved consensus (three recommendations and 14 practice points). These offer guidance on the initiation, monitoring, and evaluation of treatment with the oXiris filter in patients with sepsis, septic shock, severe COVID-19, or other hyperinflammatory states. We also recommend using the oXiris filter in patients undergoing cardiopulmonary bypass for cardiac surgery who are at risk for developing AKI. Other statements discuss modality, dosage, filter lifespan, anticoagulation, and combining the oXiris filter set with other extracorporeal therapies. Taken together, our statements offer targeted guidance for clinicians on the use of the oXiris filter across different patient populations. Additionally, we highlight the limitations of the current evidence base and identify key areas requiring further research.

Background:The study aimed to develop multiple machine learning models to predict acute kidney injury after noncardiac surgery (NCS-AKI) and screen for the best predictive model. Method:A total of 778 patients admitted to the surgical intensive care unit (SICU) were screened from July 2022 to December 2022. Six machine algorithms were used to predict early postoperative AKI. Results:A total of 704 patients were enrolled, 191 (27.1%) presented with early AKI. The survival rate of the AKI patients within 28 days was lower than that of the non-AKI (87.9% vs. 96.3%, log-rank test: P < 0.001). Among the six prediction models, the support vector machine (SVM) model had the best prediction performance, with an AUC value of 0.81 for the test set and the best net benefit compared with the other tools. The calibration curve of the model had a P-value of 0.68, indicating good fitness. The decision curve of the model indicates that over 80% of the model's predictions resulted in benefits. Conclusion:NCS-AKI had a high incidence in the SICU, and AKI patients had a lower 28-day survival rate than non-AKI patients do. The SVM had a better ability to predict early postoperative AKI.

INTRODUCTION:Septic shock-associated acute kidney injury (AKI) is known to carry a poor prognosis. The present study aims to evaluate the effect of absorptive blood purification on haemodynamic status and clinical outcomes in patients with septic shock-associated AKI. METHODS AND ANALYSIS:This multicentre, randomised controlled trial will evaluate an adsorbent filter (oXiris) in patients with abdominal septic shock requiring continuous renal replacement therapy (CRRT). The study plans to recruit 192 participants from intensive care units, who will be randomly assigned to either the intervention group (oXiris set) or the control group (ST 150 set). The primary endpoint is the reduction in norepinephrine equivalence following the initiation of CRRT. Key secondary endpoints include changes in cardiac index and systemic vascular resistance index, as measured by pulse indicator continuous cardiac output, following CRRT initiation, as well as overall survival and kidney recovery rates. ETHICS AND DISSEMINATION:The study will be conducted in accordance with the Declaration of Helsinki and Good Clinical Practice. The Ethics Committee of Shanghai Jiaotong University School of Medicine, Renji Hospital (no. LY-2024-082-B) approved this study on 27 May 2024. The results will be published in peer-reviewed journals and presented at international conferences. TRIAL REGISTRATION NUMBER:NCT06504316 (https://clinicaltrials.gov/study/NCT06504316).

This study investigated the application of software-based data monitoring for quality control (QC) in continuous renal replacement therapy (CRRT) management. This single-center pre-post intervention study, conducted in three ICUs of a tertiary hospital in Shanghai, compared outcomes before (Jan–Dec 2023) and after (Jan–Dec 2024) implementing the Sharesource Connect system. Data from 9 Prismaflex CRRT machines were collected retrospectively during 2023 and prospectively on a monthly basis during 2024. Alongside the software, a comprehensive quality improvement program: (1) multidisciplinary team collaboration; (2) data-driven QC; and (3) structured training. Primary outcomes—filter lifespan, downtime percentage, delivered/prescribed dose ratio, ultrafiltration volume, and vascular access alarms—were compared. A total of 798 filters from 514 patients (2023) and 717 filters from 492 patients (2024) were analyzed. Key quality metrics improved significantly following implementation (2024 vs. 2023): Filter lifespan increased significantly from 20.08 ± 4.12 h to 24.08 ± 4.27 h (P = 0.043), Kaplan–Meier analysis demonstrated improved filter survival (Log-Rank p < 0.001). Cumulative survival increased from 2023 to 2024 at key time points: 12 h (69.1

Background:Esophageal cancer is a common malignant tumor of the digestive tract, for which esophagectomy, a high-risk surgery, is the main treatment. Certain features of esophageal cancer surgery may increase the uncertainty of postoperative extubation, and inappropriate timing for extubation after esophagectomy may result in serious complications. This study aimed to determine the incidence and risk factors of delayed extubation after esophagectomy in the surgical intensive care unit (SICU). Methods:The study population included patients with esophageal cancer who were admitted to the SICU after undergoing transthoracic esophagectomy in one center from July 2023 to December 2023. In the SICU, patients who were extubated within 90 minutes were assigned to the early-extubation group, and those extubated after more than 90 minutes were assigned to the delayed-extubation group. The demographic characteristics, preoperative variables, intraoperative parameters, and postoperative variables were collected. Results:Of the 117 enrolled patients, there were 49 (41.9%) patients in delayed-extubation group, and compared to those in the early-extubation group, these patients had a longer operative duration (254.41±76.41 vs. 229.32±49.14 min; P=0.03), median higher colloid infusion volume {750 [interquartile range (IQR), 450-750] vs. 500 (IQR, 250-750) mL; P=0.01}, and a lower urine volume (1.79±0.9 vs. 2.36±1.41 mL/kg/h; P=0.01). Five patients received high-flow nasal cannula therapy sequentially after extubation, representing a higher proportion in the delayed-extubation group than that in the early-extubation group (10.2% vs. 1.5%; P=0.04). Compared to the early group, the delayed group had a higher hospitalization costs [10.74±3.15 vs. 9.8±1.46 (in CNY ¥10,000); P=0.04]. In the multivariate logistic regression analysis, the independent influencing factors for delayed extubation were intraoperative colloid infusion >500 mL [odds ratio (OR) =2.97; P=0.009] and per 1 mL/kg/h increasing in urine volume (OR =0.62; P=0.03). Conclusions:Delayed extubation after esophagectomy was found to be common in the SICU and remained an important factor contributing to the increase in the incidence of adverse events and medical costs. Intraoperative colloid infusion and urine volume were associated with delayed extubation.

is recommended for patients with COVID-19. However, the increasing use of HFNC brings a risk of delayed intubation. The optimal timing of switching from HFNC to invasive mechanical ventilation (IMV) remains unclear. An effective predictor is needed to assist in deciding on the timing of intubation. Respiratory rate and oxygenation (ROX) index, defined as (SpO(2) /FiO(2)) / respiratory rate, has already shown good diagnostic accuracy. Modified ROX (mROX) index, defined as (PaO2 /FiO(2)) / respiratory rate, might be better than the ROX index in predicting HFNC failure. Objective: The aim was to evaluate the predictive value of mROX for HFNC failure in patients with COVID-19. Methods: Severe or critical patients with COVID-19 treated with HFNC were enrolled in two clinical centers. Laboratory indicators, respiratory parameters, and mROX index at 0 h and 2 h after initial HFNC were collected. Based on the need for IMV after HFNC initiation, the patients were divided into an HFNC failure group and an HFNC success group. The predictive value of mROX index for IMV was evaluated by the area under the receiver operating characteristic curve (AUROC) and logistic regression analysis. We performed Kaplan-Meier survival analysis using the logrank test. Results: Sixty patients with COVID-19 (mean +/- SD age, 62.8 +/- 14.1 years; 42 patients were male) receiving HFNC were evaluated, including 18 critical and 42 severe cases. A total of 33 patients had hypertension; 14 had diabetes; 17 had chronic cardiac disease; 11 had chronic lung disease; 13 had chronic kidney disease; and 17 had a history of stroke. The AUROC of mROX index at 2 h was superior to that of other respiratory parameters to predict the need for IMV (0.959; p < 0.001). At the mROX index cutoff point of 4.45, predicting HFNC failure reached the optimal threshold, with specificity of 94% and sensitivity of 92%. Logistic regression analysis showed that 2-h mROX index < 4.45 was a protective factor for IMV (odd radio 0.18; 95% CI 0.05-0.64; p = 0.008). In the HFNC failure group, the median time from HFNC to IMV was 22.5 h. The 28-day mortality of the late intubation patients (>= 22.5 h) was higher than that of the early intubation patients ( < 22.5 h) (53.8% vs. 8.3%; p = 0.023). Conclusions: mROX at 2 h is a good early warning index of the need for IMV in patients with COVID-19 after HFNC initiation. Early intubation may lead to better survival in patients with 2-h mROX index < 4.45. (c) 2023 Elsevier Inc. All rights reserved.

There are no universally accepted indications to initiate renal replacement therapy (RRT) among patients with acute kidney injury (AKI). This study aimed to develop a nomogram to predict the risk of RRT among AKI patients in intensive care unit (ICU). In this retrospective cohort study, we extracted AKI patients from Medical Information Mart for Intensive Care III (MIMIC-III) database. Patients were randomly divided into a training cohort (70

Background: Little is known about the characteristics of lymphocyte subsets and the association with patient outcomes in COVID-19 with and without impaired kidney function. Methods: Lymphocyte subsets were compared in COVID-19 patients with or without kidney dysfunction. The primary outcome was a composite of all-cause mortality or intensive care unit admission. Secondary outcomes included duration of viral shedding, length of hospital stay, and acute kidney injury. Results: Lymphocyte subset cell counts demonstrated the lowest in patients with severe/critical COVID-19 and kidney dysfunction. Among all lymphocyte subset parameters, Th cell count was the most significant indicator for outcomes. ROC of the combined model of Th cell count and eGFR presented better predictive value than that of the other parameters. Th cell count <394.5 cells/mu l and eGFR <87.5 ml/min/1 center dot 73m(2) were independently associated with poor outcomes. The propensity score matching analysis revealed consistent results. Conclusions: Reduced Th cell count and eGFR may be applied as promising predictive indicators for identifying COVID-19 patients with high risk and poor outcomes.

BACKGROUND:COVID-19 causes significant mortality during the recent pandemic. Data regarding the effectiveness of Paxlovid on COVID-19 patients with chronic kidney disease (CKD, eGFR <90 ml/min) are limited. METHODS:A retrospective cohort study was performed on the clinical data of the hospitalized adult patients with confirmed COVID-19 infection collected at Renji Hospital from April 7, 2022 to June 21, 2022. The association of Paxlovid treatment with early (within 5 days post diagnosis) or late (5 days or later post diagnosis) initiation time with clinical outcomes was assessed by Cox proportional hazards regression model with time-dependent covariates. RESULT:1279 of 2387 enrollees were included in the study. Patients with early initiation of Paxlovid had a lower all-cause death rate compared to those with late initiation or without Paxlovid treatment (P = 0.046). For the CKD patients with Charlson comorbidity index (CCI) > 7, the early initiation of Paxlovid was associated with a lower all-cause death rate compared to the later initiation or the lack of Paxlovid treatment (P = 0.041). Cox regression analyses revealed that eGFR (HR 4.21 [95%, CI 1.62-10.99]), Paxlovid treatment (0.32 [0.13-0.77]), CCI (4.32 [1.64-11.40]), ICU admission (2.65 [1.09-6.49]), hsCRP (3.88 [1.46-7.80]), chronic liver disease (4.02 [1.09-14.85]) were the independent risk factors for all-cause death for CKD patients after adjusting for demographics and biochemical indexes. CONCLUSIONS:All-cause death, invasive ventilation, and ICU admission were all significantly lowered by an early initiation of Paxlovid treatment in COVID-19 patients with severe CKD.

Background and Objective: Intracerebral hemorrhage (ICH) is the most lethal subtype of stroke with a poor prognosis. The characteristics of ICH-related gut microbiota dysbiosis are unknown. The current study aims to investigate the differences in gut microbiota profiles between ICH patients and healthy controls. Methods: From July 8, 2019, to December 31, 2020, we performed a prospective clinical trial (Chinese Clinical Trail Registry, ChiCTR2000034906) and enrolled eligible participants, including ICH patients and healthy controls. Based on the brain computed tomography scan results, ICH patients were further categorized into thalamus hemorrhage (TH) and non-thalamus hemorrhage (NTH) subgroups. All participants were fed with the same principle of nutritional schemes, and their stool samples were collected after 5 days. The gut microbiota profiles in stool samples were determined using 16S rDNA amplicon sequencing. Clinical outcomes of the ICH patients were recorded, including the length of hospital stay, length of intensive care unit stay, neurological functions, and mortality. Results: The ICH and healthy control groups had 40 and 47 participants, respectively. The diversity (decreased Simpson and Shannon index, p = 0.002 and 0.001, respectively), microbial community structure, phylogenetic diversity, and abundant gut microbiota taxa in the ICH group were significantly different from those in the control group. However, there was no difference in gut microbiota changes and clinical outcomes between the two ICH subgroups (TH, N = 11, vs NTH, N = 29). Conclusions: Compositional alterations in gut microbiota were common in adult patients with ICH, but there was no significant difference between the ICH subtypes.

As one of the frequent complications leading to poor prognosis in hospitalized COVID-19 patients, a better understanding of venous thromboembolism (VTE) in COVID-19 patients is needed. We conducted a single-center, retrospective study on 96 COVID-19 patients admitted to the intensive care unit (ICU) from April to June 2022, in Shanghai Renji Hospital. Records of these COVID-19 patients upon admission were reviewed for demographic information, co-morbidities, vaccinations, treatment, and laboratory tests. VTE occurred in 11 (11.5%) cases among 96 COVID-19 patients despite the standard thromboprophylaxis since ICU admission. In COVID-VTE patients, a significant increase in B cells and a decrease in Ts cells were observed and a strong negative correlation (r = -0.9524, P = .0003) was found between these two populations. In COVID-19 patients with VTE, increased MPV and decreased albumin levels were seen in addition to the common VTE indicators of D-dimer abnormalities. The altered lymphocyte composition in COVID-VTE patients is noteworthy. In addition to D-dimer, MPV and albumin levels might be novel indicators for the risk of VTE in COVID-19 patients.

Alzheimer’s disease (AD) is an age-related and progressive neurodegenerative disease. In this study, we generated an induced pluripotent stem cell (iPSC) line from the dermal fibroblasts of a 69-year-old female patient carrying APOEε3/ε4 allele and diagnosed with sporadic AD. The iPSC line will be a useful tool for investigating the pathogenesis mechanisms and for drug tests in AD.