IntroductionThis study conducts a bibliometric analysis to map the intellectual structure, evolution, and emerging trends in research on airway pressure-based indexes for monitoring inspiratory effort.MethodsSystematic searches of the Web of Science Core Collection (WOSCC) and Pubmed were performed for publications dated between 1990 and 2025. Bibliometric parameters, including publication trends, country and affiliation contributions, author influence, journal distribution, keyword co-occurrence, and reference co-citation networks, were analyzed using Bibliometrix and CiteSpace.ResultsThe analysis included 291 publications from WOSCC. The annual publication output showed a near U-shaped trend, with an initial decline after the 1990s, followed by a strong resurgence after 2011. Italy was the most productive country, followed by the USA and France. The Institut National de la Sante et de la Recherche Medicale emerged as the leading institution. The journal Chest published the most articles, while the American Journal of Respiratory and Critical Care Medicine had the highest total citations. Laurent Brochard was identified as the most prolific and influential author. Keyword analysis highlighted “occlusion pressure” and “mechanical ventilation” as core themes. Reference co-citation clustering revealed major research domains, including “acute respiratory distress syndrome,” “self-inflicted lung injury,” and “nasal high flow.” Burst detection analysis indicated that “respiratory drive,” “lung injury,” and “critically ill patients” are emerging research frontiers. Complementary analysis of 242 PubMed clinical studies confirmed these trends and highlighted growing clinical focus on “fluid responsiveness” and “amyotrophic lateral sclerosis.”ConclusionResearch on airway pressure-based indices has evolved from physiological studies into a crucial clinical tool for respiratory monitoring. The field exhibits strong international collaboration and emphasizes core areas, including acute respiratory failure and lung-protective ventilation. Analysis of clinical study data confirms these trends and highlights emerging applications in the assessment of fluid responsiveness and neuromuscular disorders. These findings support the ongoing development of personalized ventilation strategies based on monitoring respiratory effort.

Background: The flow index (FI) is a noninvasive method for monitoring inspiratory effort during pressure support ventilation (PSV). However, the impact of different sampling frequencies on FI accuracy remains unclear. This study investigates how sampling frequency affects FI measurements and its diagnostic efficacy in identifying low or high inspiratory effort. Methods: A retrospective analysis of prospectively collected data from patients with acute brain injury (ABI) and acute respiratory failure (ARF) undergoing PSV with esophageal pressure monitoring was conducted. Flow, airway pressure, and esophageal pressure data were collected at 100 (ABI) and 200 Hz (ARF), then downsampled to 100, 50, 25, and 20 Hz. FI was calculated using a Python script, and respiratory mechanics were analyzed. Bland-Altman analysis compared FI at lower frequencies to the 100 Hz reference (both ABI and ARF cohorts). Receiver operating characteristic (ROC) curve analysis was conducted to evaluate the diagnostic accuracy of FI in identifying low or high inspiratory effort across sampling frequencies. Results: A total of 178 records from 76 subjects were analyzed. FI values were significantly lower at 50, 25, and 20 Hz compared with 100 Hz (P < .001). Bland-Altman analysis revealed biases of -0.08 (95% CI -0.17 to 0.00), -0.20 (95% CI -0.36 to -0.04), and -0.23 (95% CI -0.39 to -0.06) for 50, 25, and 20 Hz, respectively. The corresponding limits of agreement all exceeded the predefined clinically acceptable range. Despite these differences, the diagnostic efficacy of FI in identifying low or high inspiratory effort remained consistent across all frequencies, with area under the ROC ranging from 0.84 to 0.85 and 0.85 to 0.90, respectively. Conclusions: Although FI demonstrated robust diagnostic performance for low or high inspiratory effort across sampling frequencies, its reduced accuracy at lower frequencies calls for cautious use and frequency-specific cutoffs for effective FI monitoring.

Background Publicly accessible critical care–related databases contain enormous clinical data, but their utilization often requires advanced programming skills. The growing complexity of large databases and unstructured data presents challenges for clinicians who need programming or data analysis expertise to utilize these systems directly. Objective This study aims to simplify critical care–related database deployment and extraction via large language models. Methods The development of this platform was a 2-step process. First, we enabled automated database deployment using Docker container technology, with incorporated web-based analytics interfaces Metabase and Superset. Second, we developed the intensive care unit–generative pretrained transformer (ICU-GPT), a large language model fine-tuned on intensive care unit (ICU) data that integrated LangChain and Microsoft AutoGen. Results The automated deployment platform was designed with user-friendliness in mind, enabling clinicians to deploy 1 or multiple databases in local, cloud, or remote environments without the need for manual setup. After successfully overcoming GPT’s token limit and supporting multischema data, ICU-GPT could generate Structured Query Language (SQL) queries and extract insights from ICU datasets based on request input. A front-end user interface was developed for clinicians to achieve code-free SQL generation on the web-based client. Conclusions By harnessing the power of our automated deployment platform and ICU-GPT model, clinicians are empowered to easily visualize, extract, and arrange critical care–related databases more efficiently and flexibly than manual methods. Our research could decrease the time and effort spent on complex bioinformatics methods and advance clinical research.

OBJECTIVE:To investigate the frequency and related factors of ineffective triggering (IT) and double triggering (DT) in patients with acute brain injury undergoing invasive mechanical ventilation. METHODS:A retrospective cohort study was conducted using data from a single-center observational trial. Patients with acute brain injury [traumatic brain injury, stroke, and post-craniotomy for brain tumors] undergoing mechanical ventilation in the intensive care unit (ICU) of Beijing Tiantan Hospital, Capital Medical University between June 2017 and July 2019 were retrospectively analyzed. Demographic and clinical data were collected. Respiratory parameters and waveforms during the first 3 days of mechanical ventilation were recorded, with 15-minute waveform segments collected 4 times daily. Airway occlusion pressure (P0.1) was measured via end-expiratory hold at the end of each recording. IT and DT were identified based on airway pressure, flow, and esophageal pressure waveforms, and the ineffective triggering index (ITI) and DT incidence were calculated. Multivariate Logistic regression was used to identify factors associated with IT and DT. RESULTS:A total of 94 patients with acute brain injury were ultimately enrolled, including 19 cases of traumatic brain injury (20.2%), 39 cases of stroke (41.5%), and 36 cases of post-craniotomy for brain tumor (38.3%). Supratentorial injury was observed in 49 patients (52.1%), while infratentorial injury was identified in 45 patients (47.9%). A total of 94 patients with 1 018 datasets were analyzed; 684 (67.2%) datasets were on pressure support ventilation (PSV), and 334 (32.8%) were on mandatory ventilation. IT was detected in 810 (79.6%) datasets, with a median incidence of 2.1% (0.3%, 12.0%). Datasets demonstrating IT were characterized by lower P0.1, higher tidal volume (VT), reduced respiratory rate (RR), and decreased minute ventilation (MV) compared to those without IT. The proportion of datasets exhibiting IT was higher during PSV than in mandatory ventilation [83.8% (573/684) vs. 71.0% (237/334), P < 0.05], while, the prevalence of ITI ≥ 10% was lower [23.8% (163/684) vs. 33.5% (112/334), P < 0.05]. DT was detected in 305 datasets (30%), with a median incidence of 0.6% (0.4%, 1.3%). Datasets exhibiting DT were characterized by higher VT, reduced RR, and lower pressure support levels. The incidence of DT was lower in PSV compared to mandatory ventilation modes [0% (0%, 0.3%) vs. 0% (0%, 0.5%), P < 0.05]. The post-craniotomy for brain tumors group exhibited higher ITI, lower RR, reduced MV, and a greater proportion of infratentorial lesions, compared to the TBI group. The infratentorial lesion group demonstrated higher ITI and incidence of DT compared to the supratentorial lesion group [ITI: 3.1% (0.7%, 17.8%) vs. 1.5% (0%, 8.3%), incidence of DT: 0% (0%, 0.5%) vs. 0% (0%, 0%), both P < 0.05]. After adjusting for confounding factors through multivariate logistic regression analysis, infratentorial lesion [odds ratio (OR) = 2.029, 95% confidence interval (95%CI) was 1.465-2.811, P < 0.001], lower P0.1 (OR = 0.714, 95%CI was 0.616-0.827, P < 0.001), and mandatory ventilation (OR = 1.613, 95%CI was 1.164-2.236, P = 0.004) were independently associated with IT. Additionally, infratentorial lesion (OR = 1.618, 95%CI was 1.213-2.157, P = 0.001), large tidal volume (OR = 1.222, 95%CI was 1.137-1.314, P < 0.001), lower pressure support levels (OR = 0.876, 95%CI was 0.829-0.925, P < 0.001), and mandatory ventilation (OR = 2.750, 95%CI was 1.983-3.814, P < 0.001) were independently associated with DT. CONCLUSION:IT and DT were common in patients with acute brain injury. Infratentorial lesions and mandatory ventilation were independently associated with both IT and DT.

Background:Reverse triggering (RT) is a ventilatory asynchrony characterized by the activation of respiratory muscles in response to passive mechanical insufflation. Although RT can potentially exacerbate lung injury, its characteristics in patients with acute brain injury remain under-explored. This study aims to elucidate the incidence and factors associated with RT in this patient population. Methods:A retrospective analysis was conducted using a clinical database dedicated to investigating patient-ventilator asynchrony among individuals with acute brain injury. Only patients under controlled mechanical ventilation were included. RT was identified through an analysis of airway pressure, flow, and esophageal pressure waveforms collected at 15-minute intervals. The incidence of RT was determined by calculating the ratio of the number of RT breaths to the total number of breaths. Results:A total of 334 waveform datasets from 53 patients were analyzed. RT was observed in 8.4% of mechanical insufflations across 59 datasets (17.7%). Sixteen patients (30.2%) experienced at least one RT event. The most prevalent phenotype was mid-cycle RT (61.1%), followed by breath stacking (BD) (16.6%). Independent predictors of RT, after adjusting for confounding factors, included the combined use of opioids and sedatives, lower Sedation-Agitation Scale (SAS) scores, reduced airway delta pressure, and minimal discrepancies between the set respiratory rate and the actual respiratory rate. The pressure of occlusion at 0.1 seconds (P0.1) demonstrated substantial predictive ability for BD, with an area under the receiver operating characteristics curve of 0.72 (95% confidence interval: 0.64-0.80, P<0.001); the optimal cutoff was determined to be 1.69 cmH2O, achieving 83.3% sensitivity and 67.1% specificity. Conclusions:Factors such as deep sedation, lower airway delta pressure, and close alignment of ventilator and patient respiratory rates were associated with RT in patients with acute brain injury. Additionally, P0.1 served as a reliable predictor for the occurrence of BD.

BackgroundThe successful implementation of assisted ventilation depends on matching the patient’s effort with the ventilator support. Pressure muscle index (PMI), an airway pressure based measurement, has been used as noninvasive monitoring to assess the patient’s inspiratory effort. The authors aimed to evaluate the feasibility of pressure support adjustment according to the PMI target and the diagnostic performance of PMI to predict the contribution of the patient’s effort during ventilator support.MethodsIn this prospective physiological study, 22 adult patients undergoing pressure support ventilation were enrolled. After an end-inspiratory airway occlusion, airway pressure reached a plateau, and the magnitude of change in plateau from peak airway pressure was defined as PMI. Pressure support was adjusted to obtain the PMI which was closest to −1, 0, +1, +2, and + 3 cm H2O. Each pressure support level was maintained for 20 min. Esophageal pressure was monitored. Pressure–time products of respiratory muscle and ventilator insufflation were measured, and the fraction of pressure generated by the patient was calculated to represent the contribution of the patient’s inspiratory effort.ResultsA total of 105 datasets were collected at different PMI-targeted pressure support levels. The differences in PMI between the target and the obtained value were all within ±1 cm H2O. As targeted PMI increased, pressure support settings decreased significantly from a median (interquartile range) of 11 (10–12) to 5 (4–6) cm H2O (p < 0.001), which resulted in a significant increase in pressure–time products of respiratory muscle [from 2.9 (2.1–5.0) to 6.8 (5.3–8.1) cm H2O•s] and the fraction of pressure generated by the patient [from 25% (19–31%) to 72% (62–87%)] (p < 0.001). The area under receiver operating characteristic curves for PMI to predict 30 and 70% contribution of patient’s effort were 0.93 and 0.95, respectively. High sensitivity (all 1.00), specificity (0.86 and 0.78), and negative predictive value (all 1.00), but low positive predictive value (0.61 and 0.43) were obtained to predict either high or low contribution of patient’s effort.ConclusionOur results preliminarily suggested the feasibility of pressure support adjustment according to the PMI target from the ventilator screen. PMI could reliably predict the high and low contribution of a patient’s effort during assisted ventilation.Clinical trial registration: ClinicalTrials.gov, identifier NCT05970393.

BackgroundAcute respiratory distress syndrome (ARDS) is a severe condition characterized by lung stiffness and compromised gas exchange, often requiring mechanical ventilation for treatment. In addition to its clinical significance, understanding the publication trends and research patterns in respiratory mechanics related to ARDS can provide insights into the evolution of this field from a bibliometric perspective, aiding in strategic planning and resource allocation for future research endeavors.ObjectiveThis study aimed to explore the trends and identify the hotspots in respiratory mechanics research related to ARDS.MethodsAll relevant studies on respiratory mechanics of ARDS published between 1985 and 2023 were retrieved from the Web of Science Core Collection (WoSCC), and the retrieval strategy was topic search “TS = respiratory mechanics OR lung mechanics AND TS = ARDS OR acute respiratory distress syndrome.” Annual trends, citation patterns, and contributions from countries, institutions, authors, and journals were analyzed using Bibliometrix Biblioshiny. Networks and overlay of authors, institutions, countries, journals, co-citations, and keywords were analyzed and visualized using VOSviewer.ResultsOur analysis included 1,248 articles published between 1985 and 2023, revealing fluctuations in publication output over time. The United States emerged as the leading contributor, with Critical Care Medicine being the most prominent journal. Key research themes included mechanical ventilation, acute lung injury, and protective ventilation strategies. International collaboration was evident, facilitating knowledge exchange and interdisciplinary cooperation.ConclusionOur study sheds light on the evolving landscape of respiratory mechanics research in ARDS. International collaboration is pivotal in advancing the field, while researchers increasingly focus on personalized approaches to address the complexities of ARDS respiratory mechanics.

BACKGROUND:There is no widely accepted consensus on the weaning and extubating protocols for neurosurgical patients, leading to heterogeneity in clinical practices and high rates of delayed extubation and extubation failure-related health complications. METHODS:In this single-center prospective observational diagnostic study, mechanically ventilated neurosurgical patients with extubation attempts were consecutively enrolled for 1 yr. Responsive physicians were surveyed for the reasons for delayed extubation and developed the Swallowing, Tongue protrusion, Airway protection reflected by spontaneous and suctioning cough, and Glasgow Coma Scale Evaluation (STAGE) score to predict the extubation success for neurosurgical patients already meeting other general extubation criteria. RESULTS:A total of 3,171 patients were screened consecutively, and 226 patients were enrolled in this study. The rates of delayed extubation and extubation failure were 25% (57 of 226) and 19% (43 of 226), respectively. The most common reasons for the extubation delay were weak airway-protecting function and poor consciousness. The area under the receiver operating characteristics curve of the total STAGE score associated with extubation success was 0.72 (95% CI, 0.64 to 0.79). Guided by the highest Youden index, the cutoff point for the STAGE score was set at 6 with 59% (95% CI, 51 to 66%) sensitivity, 74% (95% CI, 59 to 86%) specificity, 90% (95% CI, 84 to 95%) positive predictive value, and 30% (95% CI, 21 to 39%) negative predictive value. At STAGE scores of 9 or higher, the model exhibited a 100% (95% CI, 90 to 100%) specificity and 100% (95% CI, 72 to 100%) positive predictive value for predicting extubation success. CONCLUSIONS:After a survey of the reasons for delayed extubation, the STAGE scoring system was developed to better predict the extubation success rate. This scoring system has promising potential in predicting extubation readiness and may help clinicians avoid delayed extubation and failed extubation-related health complications in neurosurgical patients. EDITOR’S PERSPECTIVE:

ABSTRACTCentral nervous system infections (CNSIs) are common complications after neurosurgery with a poor prognosis. The traditional microbiological culture methodology has a low detection rate and time consuming. Metagenomic next-generation sequencing (mNGS) has demonstrated the advantages of being faster, more accurate, and more comprehensive in clinical microbiology. Previous studies had suggested that mNGS had a high sensitivity in the diagnosis of CNSIs. Whether the application of mNGS has health economic value in clinical applications remains to be studied. We designed a prospective, single-center, superiority randomized controlled trial to compare the cost-effectiveness of mNGS with traditional methods for diagnosing CNSIs using a decision tree model. A total of 204 patients will be enrolled and randomly assigned to either the mNGS group or the traditional method group. The two groups of patients entered different decision points according to different clinical manifestations and examination results. They will be then given treatment decisions by a panel of specialists at the corresponding decision point. The primary outcome is the incremental cost-effectiveness ratio, which is the increased cost for every 1% increase in recovery rate. The secondary outcomes are a comparison of time cost, detection cost, and costs associated with antibiotics treatment between the two groups.IMPORTANCEDiagnosing and treating postoperative central nervous system infections (PCNSIs) remains challenging due to the low detection rate and time-consuming nature of traditional methods for identifying microorganisms in cerebrospinal fluid. Metagenomic next-generation sequencing (mNGS) technology provides a rapid and comprehensive understanding of microbial composition in PCNSIs by swiftly sequencing and analyzing the microbial genome. The current study aimed to assess the economic impact of using mNGS versus traditional bacterial culture-directed PCNSIs diagnosis and therapy in post-neurosurgical patients from Beijing Tiantan Hospital. mNGS is a relatively expensive test item, and whether it has the corresponding health-economic significance in the clinical application of diagnosing intracranial infection has not been studied clearly. Therefore, the investigators hope to explore the clinical application value of mNGS detection in PCNSIs after neurosurgery.

BACKGROUND: Ineffective effort (IE) is a frequent patient-ventilator asynchrony in invasive mechanical ventilation. This study aimed to investigate the incidence of IE and to explore its relationship with respiratory drive in subjects with acute brain injury undergoing invasive mechanical ventilation. METHODS: We retrospectively analyzed a clinical database that assessed patient-ventilator asynchrony in subjects with acute brain injury. IE was identified based on airway pressure, flow, and esophageal pressure waveforms collected at 15-min intervals 4 times daily. At the end of each data set recording, airway-occlusion pressure (P0.1) was determined by the airway occlusion test. IE index was calculated to indicate the severity of IE. The incidence of IE in different types of brain injuries as well as its relationship with P0.1 was determined. RESULTS: We analyzed 852 data sets of 71 subjects with P0.1 measured and undergoing mechanical ventilation for at least 3 d after enrollment. IE was detected in 688 (80.8%) data sets, with a median index of 2.2% (interquartile range 0.4–13.1). Severe IE (IE index ≥ 10%) was detected in 246 (28.9%) data sets. The post craniotomy for brain tumor and the stroke groups had higher median IE index and lower P0.1 compared with the traumatic brain injury group (2.6% [0.7–9.7] vs 2.7% [0.3–21] vs 1.2% [0.1–8.5], P = .002; 1.4 [1–2] cm H2O vs 1.5 [1–2.2] cm H2O vs 1.8 [1.1–2.8] cm H2O, P = .001). Low respiratory drive (P0.1 < 1.14 cm H2O) was independently associated with severe IE in the expiratory phase (IEE) even after adjusting for confounding factors by logistic regression analysis (odds ratio 5.18 [95% CI 2.69–10], P < .001). CONCLUSIONS: IE was very common in subjects with acute brain injury. Low respiratory drive was independently associated with severe IEE.

Abstract Background Assessment of the patient’s respiratory effort is essential during assisted ventilation. We aimed to evaluate the accuracy of airway pressure (P aw)-based indices to detect potential injurious inspiratory effort during pressure support (PS) ventilation. Methods In this prospective diagnostic accuracy study conducted in four ICUs in two academic hospitals, 28 adult acute respiratory failure patients undergoing PS ventilation were enrolled. A downward PS titration was conducted from 20 cmH2O to 2 cmH2O at a 2 cmH2O interval. By performing an end-expiratory airway occlusion maneuver, the negative P aw generated during the first 100 ms (P 0.1) and the maximal negative swing of P aw (∆P occ) were measured. After an end-inspiratory airway occlusion, P aw reached a plateau, and the magnitude of change in plateau from peak P aw was measured as pressure muscle index (PMI). Esophageal pressure was monitored and inspiratory muscle pressure (P mus) and P mus–time product per minute (PTPmus/min) were used as the reference standard for the patient’s effort. High and low effort was defined as P mus > 10 and < 5 cmH2O, or PTPmus/min > 200 and < 50 cmH2O s min−1, respectively. Results A total of 246 levels of PS were tested. The low inspiratory effort was diagnosed in 145 (59.0%) and 136 (55.3%) PS levels using respective P mus and PTPmus/min criterion. The receiver operating characteristic area of the three P aw-based indices by the respective two criteria ranged from 0.87 to 0.95, and balanced sensitivity (0.83–0.96), specificity (0.74–0.88), and positive (0.80–0.91) and negative predictive values (0.78–0.94) were obtained. The high effort was diagnosed in 34 (13.8%) and 17 (6.9%) support levels using P mus and PTPmus/min criterion, respectively. High receiver operating characteristic areas of the three P aw-based indices by the two criteria were found (0.93–0.95). A high sensitivity (0.80–1.00) and negative predictive value (0.97–1.00) were found with a low positive predictive value (0.23–0.64). Conclusions By performing simple airway occlusion maneuvers, the P aw-based indices could be reliably used to detect low inspiratory efforts. Non-invasive and easily accessible characteristics support their potential bedside use for avoiding over-assistance. More evaluation of their performance is required in cohorts with high effort.

Objectives To evaluate the association of tracheostomy timing with all-cause mortality in patients with mechanical ventilation (MV). Method It’s a retrospective cohort study. Adult patients undergoing invasive MV who received tracheostomy during the same hospitalization based on the Medical Information Mart for Intensive Care-III (MIMIC-III) database, were selected. The primary outcome was the relationship between tracheostomy timing and 90-day all-cause mortality. A restricted cubic spline was used to analyze the potential non-linear correlation between tracheostomy timing and 90-day all-cause mortality. The secondary outcomes included free days of MV, incidence of ventilator-associated pneumonia (VAP), free days of analgesia/sedation in the intensive care unit (ICU), length of stay (LOS) in the ICU, LOS in hospital, in-ICU mortality, and 30-day all-cause mortality. Results A total of 1,209 patients were included in this study, of these, 163 (13.5%) patients underwent tracheostomy within 4 days after intubation, while 647 (53.5%) patients underwent tracheostomy more than 11 days after intubation. The tracheotomy timing showed a U-shaped relationship with all-cause mortality, patients who underwent tracheostomy between 5 and 10 days had the lowest 90-day mortality rate compared with patients who underwent tracheostomy within 4 days and after 11 days [84 (21.1%) vs. 40 (24.5%) and 206 (31.8%), P < 0.001]. Conclusion The tracheotomy timing showed a U-shaped relationship with all-cause mortality, and the risk of mortality was lowest on day 8, but a causal relationship has not been demonstrated.

Objective:To investigate the effect of trigger sensitivity on ventilation homogeneity in patients under pressure support ventilation.Methods:We prospectively enrolled 20 patients with heterogeneous lung ventilation under pressure support ventilation that was defined by electrical impedance tomography as the distribution of tidal volume in dependent region lower than 45%. The low and high flow trigger sensitivity (the lowest and highest limits of Servo-i were 2 L/min and 0.2 L/min, respectively) were randomly applied for 20 mins. The distribution of tidal volume in dependent region and end-expiratory lung volume (EELV) were evaluated by electrical impedance tomography. The esophageal manometry was used to measure the inspiratory effort and work of breathing.Results:Comparing to the high trigger sensitivity, the low trigger sensitivity increased the relative distribution of tidal volume in dependent region [(33 ± 9)% vs. (36 ± 9)%, t = 3.735, P = 0.001], the esophageal pressure swings during inspiration [0.8 (0.4, 1.8) cmH2O vs. 1.6 (1.0, 2.1) cmH2O, Z = 2.722, P = 0.021], and pressure time product [29 (15, 54) cmH2O·s-1·min-1 vs. 48 (23, 74) cmH2O·s-1·min-1, Z = 3.298, P = 0.044]; whereas, the change of transpulmonary pressure did not significantly increase [(12.6 ± 4.3) cmH2O vs. (12.8 ± 4.2) cmH2O, t = 0.906, P = 0.376]. The global EELV of a low trigger sensitivity during pressure support ventilation was 78 (29, 170) mL, which mainly acted on the dependent region [75 (-6, 131) mL].Conclusion:Decreasing trigger sensitivity could allow more air to flow into the dependent lung region and improve homogeneity during pressure support ventilation by increasing inspiratory effort, while the working of breathing and transpulmonary pressure remain within acceptable ranges.

目的 基于半监督卷积神经网络(semi-supervised convolutional neural network,semi-CNN)构建人机不同步现象(patient-ventilator asynchrony,PVA)识别模型,评价其在压力支持通气(pressure support ventilation,PSV)模式下的诊断效能.方法 分析85例接受PSV通气脑损伤患者的机械通气数据,结合食道压监测数据进行人工标识.使用Transformer时间序列预测模型对已标识的正常或发生PVA的呼吸进行转化,转化后的数据输入semi-CNN模型判断是否发生PVA.在测试集中验证模型的准确性、灵敏度、特异度以及与专家标识结果的一致性.结果 初始训练集包含正常呼吸513次,异常呼吸69次,经过500次迭代后模型收敛.测试集包含正常呼吸48次,异常呼吸24次.在测试集中,Transformer联合semi-CNN模型识别PVA的准确率为0.92(0.83～0.97),灵敏度为0.79(0.58～0.93),特异度为0.98(0.89～1.00),Kappa值为0.80(0.65～0.95),测试结果与专家人工标识结果具有高度一致性.结论 本研究提供了一种基于semi-CNN算法的PVA识别模型,其识别PVA的准确率和特异度高,识别结果与专家人工标识结果的一致性好,可用于临床实时PVA监测.

Background The aim of the study was to determine whether the combination of Glasgow Coma Scale (GCS) and Pupil responses score (GCSP) with arterial lactate level would be an index to predict the short term prognosis in patients with traumatic brain injury (TBI). Methods A retrospective study was performed enrolling all TBI patients admitted to intensive care unit (ICU) from 2019 to 2020. The demographics, clinical characteristics, and arterial lactate concentration were recorded. The GCSP and arterial blood analysis (ABG) with lactate was tested as soon as the patient was admitted to ICU. The Glasgow Outcome Scale (GOS) after discharge was regarded as the clinical outcome. A new index named GCSP-L was the combination of GCSP and lactate concentration. GCSP-L was the GCSP score (range 1-15) plus the lactate score (range 0-2). The lactate score was defined based on different lactate concentrations. If lactate was below 2 mmol/L, lactate score was 0, which above 5 mmol/L was 2 and between 2 and 5 mmol/L, the score was 1. As the range of GCSP was 1-15, the range of the GCSP-L was 1 to 17. The area under receiver operating characteristic curve (AUC) was calculated to evaluate the predictive ability of GCSP, lactate and GCSP-L. Statistical significance was set when p value < 0.05. Results A total of 192 TBI patients were included in the study. Based on GCSP, mild, moderate, and severe TBI were 13.02, 14.06 and 72.92%, respectively. There were 103 (53.65%) patients with the lactate concentration below 2 mmol/L (1.23 ± 0.37 mmol/l), 63 (32.81%) of the range from 2 to 5 (3.04 ± 2.43 mmol/l) and 26 (13.54%) were above 5 mmol/l (7.70 ± 2.43 mmol/l). The AUC was 0.866 (95% CI 0.827-0.904) for GCSP-L, 0.812 (95% CI 0.765-0.858) for GCSP and 0.629 (95% CI 0.570—0.0.688) for lactate. The AUC of GCSP-L was higher than the other two, GCSP and lactate alone. Conclusions The combination of GCSP and lactate concentration can be used to predict the short term prognosis in TBI patients.

Objective:To investigate the relationship between bilateral frontal pneumocephalus and postoperative delirium following craniotomy for intracranial tumors.Methods:A retrospective analysis was conducted on 594 adult patients with intracranial tumors after craniotomy who were routinely monitored and treated in the Department of Critical Care Medicine, Beijing Tiantan Hospital, Capital Medical University from March to August 2018. The patients were evaluated for delirium using the Confusion Assessment Method for the ICU (CAM-ICU) on the 1st postoperative day, and the patients were divided into a delirium group and a non-delirium group according to the occurrence of postoperative delirium. Relevant clinical data before postoperative delirium assessment and outcome data at discharge were collected. According to the diagnosis of bilateral frontal pneumocephalus based on head CT within 12 hours after operation, the patients were divided into two groups according to the presence or absence of bilateral frontal pneumocephalus: bifrontal pneumocephalus group and non-bifrontal pneumocephalus group. After univariate and multivariate logistic analyses of postoperative delirium, propensity score matching (1 ∶1) was conducted to assess the relationship between bi-frontal pneumocephalus and postoperative delirium according to the influencing factors of postoperative delirium.Results:Of the 594 patients, 91 (15.3%) developed postoperative delirium. Multivariate logistic regression analysis showed that bilateral frontal pneumocephalus was an independent risk factor for postoperative delirium ( OR=1.94, 95% CI: 1.03-3.67, P=0.041). Of the 594 patients, 190 (32.0%) patients developed bilateral frontal pneumocephalus. Using the propensity score, 163 pairs of patients were successfully matched. Compared with the group without bilateral frontal pneumocephalus, the incidence of postoperative delirium was higher [11.7% (19/163) vs. 24.5% (40/163), P=0.003] and postoperative hospital stay was longer [ M ( Q1, Q3): 9 (8, 12) d vs. 11 (8, 14) d, P=0.016] in the bilateral frontal pneumocephalus group. Conclusions:Postoperative delirium has a high incidence in patients undergoing craniotomy for intracranial tumors and is associated with poor outcomes. Bilateral frontal pneumocephalus may be an important risk factor for postoperative delirium.

Abstract Background Bedside assessment of low levels of inspiratory effort, which are probably insufficient to prevent muscle atrophy, is challenging. The flow index, which is derived from the analysis of the inspiratory portion of the flow–time waveform, has been recently introduced as a non-invasive parameter to evaluate the inspiratory effort. The primary objective of the present study was to provide an external validation of the flow index to detect low inspiratory effort. Methods Datasets containing flow, airway pressure, and esophageal pressure (Pes)–time waveforms were obtained from a previously published study in 100 acute brain-injured patients undergoing pressure support ventilation. Waveforms data were analyzed offline. A low inspiratory effort was defined by one of the following criteria, work of breathing (WOB) less than 0.3 J/L, Pes–time product (PTPes) per minute less than 50 cmH2O•s/min, or inspiratory muscle pressure (Pmus) less than 5 cmH2O, adding “or occurrence of ineffective effort more than 10%” for all criteria. The flow index was calculated according to previously reported method. The association of flow index with Pes-derived parameters of effort was investigated. The diagnostic accuracy of the flow index to detect low effort was analyzed. Results Moderate correlations were found between flow index and WOB, Pmus, and PTPes per breath and per minute (Pearson’s correlation coefficients ranged from 0.546 to 0.634, P < 0.001). The incidence of low inspiratory effort was 62%, 51%, and 55% using the definition of WOB, PTPes per minute, and Pmus, respectively. The area under the receiver operating characteristic curve for flow index to diagnose low effort was 0.88, 0.81, and 0.88, for the three respective definition. By using the cutoff value of flow index less than 2.1, the diagnostic performance for the three definitions showed sensitivity of 0.95–0.96, specificity of 0.57–0.71, positive predictive value of 0.70–0.84, and negative predictive value of 0.90–0.93. Conclusions The flow index is associated with Pes-based inspiratory effort measurements. Flow index can be used as a valid instrument to screen low inspiratory effort with a high probability to exclude cases without the condition.

Objective:To observe the application of regional recruitment-to-inflation ratio (RI ratio) monitored by electrical impedance tomography (EIT) in the models of acute respiratory distress syndrome (ARDS).Methods:ARDS models were established by intravenous infusion of oleic acid in 6 Bama pigs. The model was considered as being successfully established when the ratio of the partial pressure of oxygen in arterial blood PaO2 to FiO2 (PaO2/FiO2) remained below 200 mmHg for at least 30 min. After the ARDS model was established, a recruitment manuver (RM) was performed. After RM, the animal was ventilated at PEEP 15 cmH2O for one hour. Then, a prolonged expiration was performed and PEEP was decreased from 15 to 5 cmH2O. The process was monitored by EIT, and offline analysis was performed after the experiment. Lung images were divided into two ventral-to-dorsal horizontal regions of interest (ROI). Recruitment volume (Vrec), global RI ratio, regional Vrec and regional RI ratio were calculated by EIT.Results:ARDS was successfully induced in all experimental animals with median (IQR) PaO2/FiO2 decreasing from (422.5±49.9) mmHg to (151.7±13.4) mmHg (t=14.100, P＜0.001). The median and interquartile range of the RI ratio of all experimental animals was 1.59 (1.42, 1.67). The global RI ratio, RI ratio of nondependent areas and dependent areas of 6 animals were 1.83, 1.03, 2.06; 1.68, 1.51, 2.28; 1.64, 0.98, 1.83; 1.23, 0.93, 1.33; 1.38, 0.02, 1.27; 1.55, 1.41, 1.99 respectively.Conclusion:Regional RI ratio can be calculated and monitored by EIT in ARDS patients, which may provide more information on regional recruitability.

目的 探讨平均动脉压(mean arterial pressure,MAP)变异度与重症患者的重症医学科(intensive care unit,ICU)病死率之间的关系.方法 回顾性分析重症监护医学信息数据库(Medical Information Mart for Intensive Care,MIMIC)-Ⅲ中38852例入ICU的重症患者的临床资料,计算入ICU后24 h内记录的MAP的变异系数作为MAP变异度,采用一般线性回归观察入ICU 24 h内MAP变异度与重症患者ICU病死率之间的相关性,并采用受试者工作特征(receiver operating characteristic,ROC)曲线下面积(area under curve,AUC)评估MAP变异度预测重症患者ICU病死率的能力.结果 入ICU 24 h的MAP变异度与ICU病死率之间有很好的相关性(R2=0.860,P<0.001),MAP变异度越大,ICU病死率越高.24h的MAP变异程度预测ICU病死率的AUC为0.61.结论 重症患者入ICU 24 h内的MAP变异度与ICU病死率有很好的相关性,MAP变异度越大,ICU病死率越高;MAP变异度能够为简单快速预测危重患者的ICU病死率提供一定的信息.

Objective:To compare the recruitability between pulmonary (ARDSp) and extrapulmonary acute respiratory distress syndrome (ARDSexp) using electric impedance tomography (EIT) in pig models.Methods:Sixteen healthy Bama pigs were randomly divided into ARDSp and ARDSexp group with 8 animals each. ARDSp model was established using alveolar instillation of hydrochloric acid and ARDSexp was injected with oleic acid through atrium venous. After the model was established, the animal received lung protective ventilation. After a recruitment maneuver, low (5 cmH2O) and high (25 cmH2O) PEEP were applied in each animal remaining other setting unchanged for 1 hour. Then an end-inspiratory and end-expiratory occlusion were performed. The recruitment volume (VREC) generated between high and low PEEP was calculated, and the data of oxygenation and hemodynamic were recorded.Results:Significant difference was found in VREC between ARDSexp and ARDSp models (P=0.003), and VREC was (1572.8±314.9) ml and (1089.9±224.2) ml, respectively. The ratio of arterial partial pressure of oxygen to inspired oxygen fraction (PaO2/FiO2) increased from (214.3±128.6) mmHg to (447.5±96.0) mmHg (P=0.004) with increasing PEEP in ARDSexp. However, PaO2/FiO2 did not change significantly in ARDSp [(173.6±112.0) mmHg vs (178.4±128.2) mmHg, P=0.943].Conclusion:The recruitability between ARDSp and ARDSexp in pig models is different. Compared with ARDSp, ARDSexp may have better recruitability. Higher PEEP may improve the oxygenation of ARDSexp, while the effect on ARDSp is relatively small.