Critical care explorations最新文献

{"title":"Why Has Biomarker-Guided Fluid Resuscitation for Sepsis Not Been Implemented in Clinical Practice?","authors":"Sanne Ter Horst, Jan C Ter Maaten, Matijs van Meurs, Jill Moser, Hjalmar R Bouma","doi":"10.1097/CCE.0000000000001274","DOIUrl":"10.1097/CCE.0000000000001274","url":null,"abstract":"<p><p>Sepsis is a dysregulated, potentially fatal host response to infection, characterized by heterogeneity in clinical presentation and organ failure mechanisms. Early hemodynamic resuscitation and antibiotics are crucial treatments. Current guidelines recommend a one-size-fits-all approach of 30 mL/kg fluids, which may worsen vascular leakage and organ dysfunction in some patients. Personalized strategies using biomarkers and dynamic fluid responsiveness assessments offer a more tailored approach, potentially preventing fluid overload while ensuring perfusion. A recent multiomics analysis identified sepsis subgroups benefiting from either liberal or restrictive fluid resuscitation, highlighting -omics' potential in personalized fluid management and the role of immune regulation and endothelial dysfunction in septic shock. Despite progress, methodological challenges hinder clinical implementation of biomarkers. Addressing issues like rapid point-of-care biomarker assays already at emergency department or ICU admission, standardizing sepsis diagnosis, robust external validation, and clinical trial enrichment is crucial for advancing biomarker-guided fluid management in clinical settings.</p>","PeriodicalId":93957,"journal":{"name":"Critical care explorations","volume":"7 6","pages":"e1274"},"PeriodicalIF":0.0,"publicationDate":"2025-06-09","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12151020/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144251327","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Optimizing Recovery: Early Versus Delayed Chest Tube Removal in Pediatric Cardiac Surgery Patients: A Randomized Controlled Trial.","authors":"Abdulraouf M Z Jijeh, Ghassan A Shaath, Sameh R Ismail, Mohamed S Kabbani, Omar M Hijazi, Hayan Altaweel, Husam Hamada, Ammar Qadi, Anis Fatima, Abdrabo Abdrabo, Wiaam Ahmed, Nuha Ahmed, Ahmed Elsaoudi, Ahmed Yousef, Rehana Shafi, Husam I Ardah, Ahmad Elwy, Abdullah A Alghamdi","doi":"10.1097/CCE.0000000000001271","DOIUrl":"10.1097/CCE.0000000000001271","url":null,"abstract":"<p><strong>Objectives: </strong>To evaluate the safety and efficacy of an early chest tube removal protocol in reducing tube duration without increasing complications following pediatric cardiac surgery.</p><p><strong>Design: </strong>A single-center, randomized controlled trial.</p><p><strong>Setting: </strong>Pediatric cardiac ICU.</p><p><strong>Patients: </strong>Two hundred fifteen pediatric patients with chest tubes after cardiac surgery.</p><p><strong>Interventions: </strong>Patients were randomized to early removal (drainage threshold < 6 mL/kg over 8 hr) or late removal (24-hr assessment) groups. Primary outcomes included chest tube duration, whereas secondary outcomes encompassed ICU stay, ventilation time, hospital stay, and complication rates.</p><p><strong>Measurements and main results: </strong>Median chest tube duration was significantly shorter in the early removal group (3 d) compared with the late removal group (4.9 d; p < 0.0001). Rates of fluid reaccumulation and pneumothorax were low and comparable between groups. Notably, no patients in either group required tube reinsertion. ICU and total hospital stay durations were similar across groups.</p><p><strong>Conclusions: </strong>An early chest tube removal protocol following pediatric cardiac surgery suggests a reduction in chest tube duration without increasing the risk of complications. These findings support the adoption of an evidence-based early removal approach to enhance patient comfort and optimize ICU resource utilization in pediatric cardiac surgery patients.</p>","PeriodicalId":93957,"journal":{"name":"Critical care explorations","volume":"7 6","pages":"e1271"},"PeriodicalIF":0.0,"publicationDate":"2025-06-04","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12140733/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144227984","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Retrospective Analysis of a Multicenter Observational Study of the Relationship Between Social Determinants of Health and Complications After Children's Heart Surgery.","authors":"Khurram Mustafa, Christopher Leahy, Deborah Ridout, Katherine L Brown","doi":"10.1097/CCE.0000000000001270","DOIUrl":"10.1097/CCE.0000000000001270","url":null,"abstract":"<p><strong>Importance: </strong>Epidemiological studies have highlighted disparities in illnesses and outcomes for critically unwell children.</p><p><strong>Objectives: </strong>We aimed to describe social characteristics and explore links with the outcome of postoperative complications with children's heart surgery.</p><p><strong>Design, setting, and participants: </strong>Retrospective analysis of a multicenter observational dataset including those under 17 years old undergoing heart surgery from October 2015 to June 2017 at five U.K. children's cardiac centers.</p><p><strong>Main outcomes and measures: </strong>Univariate and multivariable multinomial regression analyses were undertaken for the outcome of predefined postoperative complications.</p><p><strong>Results: </strong>Of 2898 cases meeting criteria, 2708 had complete data. Two thousand one hundred three (77.66%) had no complications, 369 (13.62%) had a single complication, 56 (2.06%) received Extracorporeal Life Support, and 179 (6.61%) had multiple complications. Children residing in low deprivation neighborhoods were under-represented: lowest quintile 361 (13.33%). Minoritized ethnic group was strongly linked to indices of deprivation: residence in neighborhoods with highest deprivation occurred with Bangladeshi, Black African, and Pakistani ethnicity and lowest deprivation with White ethnicities. Adjusted for clinical risk factors compared with the reference group (White), patients from Asian background had a significantly higher risk of developing single vs. no complications (odds ratio [OR], 1.53; 95% CI, 1.00-2.32) and Black patients had a higher risk of developing multiple vs. no complications (OR, 2.19; 95% CI, 1.09-4.41). Among single complications, Asian children had a higher risk of developing feeding issues (OR, 2.07; 95% CI, 1.13-3.28).</p><p><strong>Conclusions and relevance: </strong>Ethnicity and socioeconomic deprivation may be linked to greater risk of certain complications after pediatric cardiac surgery. Further exploration of inequities is needed in this population.</p>","PeriodicalId":93957,"journal":{"name":"Critical care explorations","volume":"7 6","pages":"e1270"},"PeriodicalIF":0.0,"publicationDate":"2025-06-04","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12140732/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144227986","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Mortality Prediction Performance Under Geographical, Temporal, and COVID-19 Pandemic Dataset Shift: External Validation of the Global Open-Source Severity of Illness Score Model.","authors":"Takeshi Tohyama, Liam G McCoy, Euma Ishii, Sahil Sood, Jesse Raffa, Takahiro Kinoshita, Leo Anthony Celi, Satoru Hashimoto","doi":"10.1097/CCE.0000000000001275","DOIUrl":"10.1097/CCE.0000000000001275","url":null,"abstract":"<p><strong>Background: </strong>Risk-prediction models are widely used for quality of care evaluations, resource management, and patient stratification in research. While established models have long been used for risk prediction, healthcare has evolved significantly, and the optimal model must be selected for evaluation in line with contemporary healthcare settings and regional considerations.</p><p><strong>Objectives: </strong>To evaluate the geographic and temporal generalizability of the models for mortality prediction in ICUs through external validation in Japan.</p><p><strong>Derivation cohort: </strong>Not applicable.</p><p><strong>Validation cohort: </strong>The care Japanese Intensive care PAtient Database from 2015 to 2022.</p><p><strong>Prediction model: </strong>The Global Open-Source Severity of Illness Score (GOSSIS-1), a modern risk model utilizing machine learning approaches, was compared with conventional models-the Acute Physiology and Chronic Health Evaluation (APACHE-II and APACHE-III)-and a locally calibrated model, the Japan Risk of Death (JROD).</p><p><strong>Results: </strong>Despite the demographic and clinical differences of the validation cohort, GOSSIS-1 maintained strong discrimination, achieving an area under the curve of 0.908, comparable to APACHE-III (0.908) and JROD (0.910). It also exhibited superior calibration, achieving a standardized mortality ratio (SMR) of 0.89 (95% CI, 0.88-0.90), significantly outperforming APACHE-II (SMR, 0.39; 95% CI, 0.39-0.40) and APACHE-III (SMR, 0.46; 95% CI, 0.46-0.47), and demonstrating a performance close to that of JROD (SMR, 0.97; 95% CI, 0.96-0.99). However, performance varied significantly across disease categories, with suboptimal calibration for neurologic conditions and trauma. While the model showed temporal stability from 2015 to 2019, performance deteriorated during the COVID-19 pandemic, broadly reducing performance across disease categories in 2020. This trend was particularly pronounced in GOSSIS compared with APACHE-III.</p><p><strong>Conclusions: </strong>GOSSIS-1 demonstrates robust discrimination despite substantial geographic dataset shift but shows important calibration variations across disease categories. In particular, in a complex model like GOSSIS-1, stresses on the health system, such as a pandemic, can manifest changes in model calibration.</p>","PeriodicalId":93957,"journal":{"name":"Critical care explorations","volume":"7 6","pages":"e1275"},"PeriodicalIF":0.0,"publicationDate":"2025-06-04","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12140679/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144227983","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Quantifying Healthcare Provider Perceptions of a Novel Deep Learning Algorithm to Predict Sepsis: Electronic Survey.","authors":"Karthik Ramesh, Aaron Boussina, Supreeth P Shashikumar, Atul Malhotra, Christopher A Longhurst, Christopher S Josef, Kimberly Quintero, Jake Del Rosso, Shamim Nemati, Gabriel Wardi","doi":"10.1097/CCE.0000000000001276","DOIUrl":"10.1097/CCE.0000000000001276","url":null,"abstract":"<p><strong>Importance: </strong>Sepsis is a major cause of morbidity and mortality, with early intervention shown to improve outcomes. Predictive modeling and artificial intelligence (AI) can aid in early sepsis recognition, but there remains a gap between algorithm development and clinical use. Despite the importance of user experience in adopting clinical predictive models, few studies have focused on provider acceptance and feedback.</p><p><strong>Objectives: </strong>To evaluate healthcare worker perception and acceptance of a deep learning sepsis prediction model in the emergency department (ED).</p><p><strong>Design, setting, and participants: </strong>COnformal Multidimensional Prediction Of SEpsis Risk (COMPOSER), a deep learning algorithm, is used at two EDs of a large academic medical center to predict sepsis before clear clinical presentation. An internally developed survey following the Checklist for Reporting Results of Internet E-Surveys was distributed to team members who received a COMPOSER alert.</p><p><strong>Analysis: </strong>Mann-Whitney U testing was performed on results stratified by provider experience.</p><p><strong>Results: </strong>A total of 114 responses were received: 76 from doctors of medicine/doctors of osteopathic medicine, 34 from registered nurses, and four from nurse practicioners/physician assistants. Of these, 53% were from providers with fewer than 5 years of experience. Seventy-seven percent of respondents had a positive or neutral perception of the alert's usefulness. Providers with 0-5 years of experience were more likely to expect sepsis after the alert (p = 0.021) and found the alert more useful (p = 0.016) compared with those with 6+ years of experience. Additionally, physicians with 0-5 years of experience were more likely to say the alert changed their patient management (p = 0.048).</p><p><strong>Conclusions: </strong>Less experienced providers were more likely to perceive benefit from the alert, which was overall received favorably. Future AI implementations might consider tailored alert patterns and education to enhance reception and reduce fatigue.</p>","PeriodicalId":93957,"journal":{"name":"Critical care explorations","volume":"7 6","pages":"e1276"},"PeriodicalIF":0.0,"publicationDate":"2025-06-04","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12140755/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144227985","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Positive End-Expiratory Pressure Titration Based on Lung Mechanics May Improve Pulse Pressure Variation Interpretation in Acute Respiratory Distress Syndrome Patients.","authors":"Vasiliki Tsolaki, George E Zakynthinos, Nikitas Karavidas, Maria Eirini Papadonta, Ilias Dimeas, Kyriaki Parisi, Theofilos Amanatidis, Epaminondas Zakynthinos","doi":"10.1097/CCE.0000000000001273","DOIUrl":"10.1097/CCE.0000000000001273","url":null,"abstract":"<p><strong>Objectives: </strong>To evaluate the effects of positive end-expiratory pressure (PEEP) on pulse pressure variation (PPV) in patients with moderate/severe acute respiratory distress syndrome (ARDS).</p><p><strong>Design: </strong>Prospective interventional self-controlled study.</p><p><strong>Setting: </strong>University Hospital of Larissa.</p><p><strong>Patients: </strong>ARDS patients admitted intubated in the ICU (from August 2020 to March 2022).</p><p><strong>Interventions: </strong>None.</p><p><strong>Measurements and main results: </strong>PPV and inferior vena cava (IVC) respiratory variability were evaluated at two PEEP levels (first value mainly based on PEEP/Fio2 and second value based on respiratory system compliance). Additionally, respiratory mechanics, hemodynamics, and echocardiographic indices assessing right ventricular (RV) size (RV end-diastolic area/left ventricular end-diastolic area [RVEDA/LVEDA]), RV systolic function, and RV afterload (pulmonary artery systolic pressure [PASP] and PASP/left ventricular outflow tract velocity time integral [PASP/VTILVOT]) were recorded. Ninety-five patients were evaluated. PPV decreased after PEEP reduction (11.7 ± 0.2 to 7.9% ± 0.2%), whereas IVC respiratory variability increased (9.1 ± 0.9 to 14.6% ± 0.1%) and central venous pressure decreased (all p < 0.0001). RV afterload indices decreased (p < 0.0001), simultaneously with RV size (< 0.0001) and systolic function indices' improvements (< 0.05); shock warranted less noradrenaline doses. The change in PPV correlated significantly to respiratory variability in IVC diameter distensibility (p < 0.0001) and moderately to changes in RV size and systolic function (change in RVEDA/change in LVEDA, change in tricuspid annular plane systolic excursion); RV afterload (change in PASP [ΔPASP], ΔPASP/VTILVOT); and change in Paco2 (all p < 0.05).</p><p><strong>Conclusions: </strong>PPV alteration with PEEP decrease, associated with IVC distensibility increases, may indicate the presence of RV dysfunction and increased pulmonary vascular resistances. Whether the patients are in need for fluid loading, fluid responsiveness assessment may be further warranted.</p>","PeriodicalId":93957,"journal":{"name":"Critical care explorations","volume":"7 6","pages":"e1273"},"PeriodicalIF":0.0,"publicationDate":"2025-05-28","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12122179/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144176177","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Predictions With a Purpose: Elevating Standards for Clinical Modeling Research.","authors":"Patrick G Lyons","doi":"10.1097/CCE.0000000000001268","DOIUrl":"10.1097/CCE.0000000000001268","url":null,"abstract":"","PeriodicalId":93957,"journal":{"name":"Critical care explorations","volume":"7 6","pages":"e1268"},"PeriodicalIF":0.0,"publicationDate":"2025-05-27","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12119043/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144176183","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Predictions With a Purpose: Elevating Standards for Clinical Modeling Research.","authors":"Patrick G Lyons","doi":"10.1097/CCE.0000000000001268","DOIUrl":"10.1097/CCE.0000000000001268","url":null,"abstract":"","PeriodicalId":93957,"journal":{"name":"Critical care explorations","volume":"7 6","pages":"e1268"},"PeriodicalIF":0.0,"publicationDate":"2025-05-27","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12119043/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144183119","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Machine Learning Accurately Predicts Need for Critical Care Support in Patients Admitted to Hospital for Community-Acquired Pneumonia.","authors":"George S Chen, Terry Lee, Jennifer L Y Tsang, Alexandra Binnie, Anne McCarthy, Juthaporn Cowan, Patrick Archambault, Francois Lellouche, Alexis F Turgeon, Jennifer Yoon, Francois Lamontagne, Allison McGeer, Josh Douglas, Peter Daley, Robert Fowler, David M Maslove, Brent W Winston, Todd C Lee, Karen C Tran, Matthew P Cheng, Donald C Vinh, John H Boyd, Keith R Walley, Joel Singer, John C Marshall, James A Russell","doi":"10.1097/CCE.0000000000001262","DOIUrl":"https://doi.org/10.1097/CCE.0000000000001262","url":null,"abstract":"<p><strong>Objectives: </strong>Hospitalized community-acquired pneumonia (CAP) patients are admitted for ventilation, vasopressors, and renal replacement therapy (RRT). This study aimed to develop a machine learning (ML) model that predicts the need for such interventions and compare its accuracy to that of logistic regression (LR).</p><p><strong>Design: </strong>This retrospective observational study trained separate models using random-forest classifier (RFC), support vector machines (SVMs), Extreme Gradient Boosting (XGBoost), and multilayer perceptron (MLP) to predict three endpoints: eventual use of invasive ventilation, vasopressors, and RRT during hospitalization. RFC-based models were overall most accurate in a derivation COVID-19 CAP cohort and were validated in one COVID-19 CAP and two non-COVID-19 CAP cohorts.</p><p><strong>Setting: </strong>This study is part of the Community-Acquired Pneumonia: Toward InnoVAtive Treatment (CAPTIVATE) Research program.</p><p><strong>Patients: </strong>Two thousand four hundred twenty COVID-19 and 1909 non-COVID-19 CAP patients over 18 years old hospitalized and not needing invasive ventilation, vasopressors, and RRT on the day of admission were included.</p><p><strong>Interventions: </strong>None.</p><p><strong>Measurements and main results: </strong>Performance was evaluated with area under the receiver operating characteristic curve (AUROC) and accuracy. RFCs performed better than XGBoost, SVM, and MLP models. For comparison, we evaluated LR models in the same cohorts. AUROC was very high ranging from 0.74 to 0.95 in predicting ventilation, vasopressors, and RRT use in our derivation and validation cohorts. ML used and variables such as Fio2, Glasgow Coma Scale, and mean arterial pressure to predict ventilator, vasopressor use, creatinine, and potassium to predict RRT use. LR was less accurate than ML, with AUROC ranging 0.66 to 0.8.</p><p><strong>Conclusions: </strong>A ML algorithm more accurately predicts need of invasive ventilation, vasopressors, or RRT in hospitalized non-COVID-19 CAP and COVID-19 patients than regression models and could augment clinician judgment for triage and care of hospitalized CAP patients.</p>","PeriodicalId":93957,"journal":{"name":"Critical care explorations","volume":"7 6","pages":"e1262"},"PeriodicalIF":0.0,"publicationDate":"2025-05-27","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144210429","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Machine Learning Accurately Predicts Need for Critical Care Support in Patients Admitted to Hospital for Community-Acquired Pneumonia.","authors":"George S Chen, Terry Lee, Jennifer L Y Tsang, Alexandra Binnie, Anne McCarthy, Juthaporn Cowan, Patrick Archambault, Francois Lellouche, Alexis F Turgeon, Jennifer Yoon, Francois Lamontagne, Allison McGeer, Josh Douglas, Peter Daley, Robert Fowler, David M Maslove, Brent W Winston, Todd C Lee, Karen C Tran, Matthew P Cheng, Donald C Vinh, John H Boyd, Keith R Walley, Joel Singer, John C Marshall, James A Russell","doi":"10.1097/CCE.0000000000001262","DOIUrl":"10.1097/CCE.0000000000001262","url":null,"abstract":"<p><strong>Objectives: </strong>Hospitalized community-acquired pneumonia (CAP) patients are admitted for ventilation, vasopressors, and renal replacement therapy (RRT). This study aimed to develop a machine learning (ML) model that predicts the need for such interventions and compare its accuracy to that of logistic regression (LR).</p><p><strong>Design: </strong>This retrospective observational study trained separate models using random-forest classifier (RFC), support vector machines (SVMs), Extreme Gradient Boosting (XGBoost), and multilayer perceptron (MLP) to predict three endpoints: eventual use of invasive ventilation, vasopressors, and RRT during hospitalization. RFC-based models were overall most accurate in a derivation COVID-19 CAP cohort and were validated in one COVID-19 CAP and two non-COVID-19 CAP cohorts.</p><p><strong>Setting: </strong>This study is part of the Community-Acquired Pneumonia: Toward InnoVAtive Treatment (CAPTIVATE) Research program.</p><p><strong>Patients: </strong>Two thousand four hundred twenty COVID-19 and 1909 non-COVID-19 CAP patients over 18 years old hospitalized and not needing invasive ventilation, vasopressors, and RRT on the day of admission were included.</p><p><strong>Interventions: </strong>None.</p><p><strong>Measurements and main results: </strong>Performance was evaluated with area under the receiver operating characteristic curve (AUROC) and accuracy. RFCs performed better than XGBoost, SVM, and MLP models. For comparison, we evaluated LR models in the same cohorts. AUROC was very high ranging from 0.74 to 0.95 in predicting ventilation, vasopressors, and RRT use in our derivation and validation cohorts. ML used and variables such as Fio₂, Glasgow Coma Scale, and mean arterial pressure to predict ventilator, vasopressor use, creatinine, and potassium to predict RRT use. LR was less accurate than ML, with AUROC ranging 0.66 to 0.8.</p><p><strong>Conclusions: </strong>A ML algorithm more accurately predicts need of invasive ventilation, vasopressors, or RRT in hospitalized non-COVID-19 CAP and COVID-19 patients than regression models and could augment clinician judgment for triage and care of hospitalized CAP patients.</p>","PeriodicalId":93957,"journal":{"name":"Critical care explorations","volume":"7 6","pages":"e1262"},"PeriodicalIF":0.0,"publicationDate":"2025-05-27","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12119046/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144183146","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}