Comparison of artificial intelligence and logistic regression models for mortality prediction in acute respiratory distress syndrome: a systematic review and meta-analysis

Intensive Care Medicine Experimental, Journal Year: 2025, Volume and Issue: 13(1)

Published: Feb. 21, 2025

Abstract Background The application of artificial intelligence (AI) in predicting the mortality acute respiratory distress syndrome (ARDS) has garnered significant attention. However, there is still a lack evidence-based support for its specific diagnostic performance. Thus, this systematic review and meta-analysis was conducted to evaluate effectiveness AI algorithms ARDS mortality. Method We comprehensive electronic search across Web Science, Embase, PubMed, Scopus , EBSCO databases up April 28, 2024. QUADAS-2 tool used assess risk bias included articles. A bivariate mixed-effects model applied meta-analysis. Sensitivity analysis, meta-regression tests heterogeneity were also performed. Results Eight studies analysis. sensitivity, specificity, summarized receiver operating characteristic (SROC) AI-based validation set 0.89 (95% CI 0.79–0.95), 0.72 0.65–0.78), 0.84 0.80–0.87), respectively. For logistic regression (LR) model, SROC 0.78 0.74–0.82), 0.68 0.60–0.76), 0.81 0.77–0.84). demonstrated superior predictive accuracy compared LR model. Notably, performed better patients with moderate severe (SAUC: [95% 0.80–0.87] vs. 0.77–0.84]). Conclusion showed performance strong potential clinical application. Additionally, we found that ARDS, highly heterogeneous condition, influenced by severity disease.

Language: Английский

---

Generation of short-term follow-up chest CT images using a latent diffusion model in COVID-19

Japanese Journal of Radiology, Journal Year: 2024, Volume and Issue: unknown

Published: Nov. 25, 2024

Abstract Purpose Despite a global decrease in the number of COVID-19 patients, early prediction clinical course for optimal patient care remains challenging. Recently, usefulness image generation medical images has been investigated. This study aimed to generate short-term follow-up chest CT using latent diffusion model patients with COVID-19. Materials and methods We retrospectively enrolled 505 whom parameters (patient background, symptoms, blood test results) upon admission were available imaging was performed. Subject datasets ( n = 505) allocated training 403), remaining 102) reserved evaluation. The underwent variational autoencoder (VAE) encoding, resulting vectors. information consisting initial radiomic features formatted as table data encoder. Initial vectors encoders utilized model. evaluation used prognostic images. Then, similarity (generated images) (real evaluated by zero-mean normalized cross-correlation (ZNCC), peak signal-to-noise ratio (PSNR), structural (SSIM). Visual assessment also performed numerical rating scale. Results Prognostic generated Image showed reasonable values 0.973 ± 0.028 ZNCC, 24.48 3.46 PSNR, 0.844 0.075 SSIM. two pulmonologists one radiologist yielded mean score. Conclusions validity predictive COVID-19-associated pneumonia reasonable. may suggest potential utility other respiratory diseases.

Language: Английский

---

Accuracy of artificial intelligence algorithms in predicting acute respiratory distress syndrome: a systematic review and meta-analysis

BMC Medical Informatics and Decision Making, Journal Year: 2025, Volume and Issue: 25(1)

Published: Jan. 28, 2025

Acute respiratory distress syndrome (ARDS) is a serious threat to human life. Hence, early and accurate diagnosis treatment are crucial for patient survival. This meta-analysis evaluates the accuracy of artificial intelligence in ARDS provides guidance future research applications. A search on PubMed, Embase, Cochrane, Web Science, CNKI, Wanfang, Chinese Biomedical Literature (CBM), VIP databases was systematically conducted, from their establishment November 2023, obtain eligible studies analysis evaluation predictive effect AI ARDS. The retrieved literature screened according inclusion exclusion criteria, quality included assessed using QUADAS-2, were statistically analyzed. Among 2, 996 studies, 33 this meta-analysis, which showed that pooled sensitivity predicting 0.81 (0.76-0.85), specificity 0.88 (0.84-0.91), area under receiver operating characteristic curve (AUC) 0.91 (0.88-0.93). analyzed 28 models, with 0.79 (0.76-0.82), 0.85 (0.83-0.88), an AUC 0.89 (0.86-0.91). In subgroup analysis, models ANN, CNN, LR, RF, SVM, XGB 0.86 (0.83-0.89), (0.88-0.93), (0.86-0.91), 0.90 (0.87-0.92), 0.93 (0.90-0.95), respectively. additional we evaluated performance trained different predictors. registered PROSPERO (CRD42023491546). has good ARDS, indicating high clinical application value. Algorithmic such as have improved prediction performance. revealed model images combined other predictors had best

Language: Английский

---

Development and External Validation of a Detection Model to Retrospectively Identify Patients With Acute Respiratory Distress Syndrome

Critical Care Medicine, Journal Year: 2025, Volume and Issue: unknown

Published: April 8, 2025

Objective: The aim of this study was to develop and externally validate a machine-learning model that retrospectively identifies patients with acute respiratory distress syndrome (acute [ARDS]) using electronic health record (EHR) data. Design: In retrospective cohort study, ARDS identified via physician-adjudication in three cohorts hypoxemic failure (training, internal validation, external validation). Machine-learning models were trained classify vital signs, support, laboratory data, medications, chest radiology reports, clinical notes. best-performing assessed internally validated the area under receiver-operating curve (AUROC), precision-recall curve, integrated calibration index (ICI), sensitivity, specificity, positive predictive value (PPV), timing. Patients: Patients undergoing mechanical ventilation within two distinct systems Interventions: None. Measurements Main Results: There 1,845 training cohort, 556 validation 199 cohort. prevalence 19%, 17%, 31%, respectively. Regularized logistic regression analyzing structured data (EHR model) reports (EHR-radiology had best performance. During EHR-radiology AUROC 0.91 (95% CI, 0.88–0.93) 0.88 0.87–0.93), Externally, ICI 0.13 0.08–0.18). At specified threshold, sensitivity specificity 80% 75%–98%), PPV 64% 58%–71%), median 2.2 hours (interquartile range 0.2–18.6) after meeting Berlin criteria. Conclusions: EHR can identify across different institutions.

Language: Английский

---

Acute Respiratory Distress Syndrome – quo vadis

Medizinische Klinik - Intensivmedizin und Notfallmedizin, Journal Year: 2025, Volume and Issue: unknown

Published: April 22, 2025

Zusammenfassung Das akute Atemnotsyndrom (ARDS) ist ein heterogenes klinisches Syndrom, das sich durch eine variable Pathophysiologie und unterschiedliche therapeutische Ansätze auszeichnet. Die jüngsten Leitlinien betonen die Bedeutung der Bauchlagerung venovenösen extrakorporalen Membranoxygenierung (vv-ECMO) für schwerste Fälle, während routinemäßige Recruitmentmanöver extrakorporale CO 2 -Eliminationsverfahren nicht mehr empfohlen werden. Um Personalisierung ARDS-Therapie weiter voranzutreiben, zeigt Identifikation von ARDS-Phänotypen mittels „latent class analysis“ vielversprechende zur personalisierten Therapie. Zudem könnten adaptive Plattformstudien auf künstlicher Intelligenz (KI) basierende Entscheidungsunterstützungssysteme ARDS-Behandlung optimieren. zukünftige wird zunehmend individualisiert sein einer verbesserten Patientenstratifizierung, neuen Studiendesigns dem gezielten Einsatz moderner Technologien basieren. Dieser Artikel fasst aktuellen Entwicklungen in zusammen, insbesondere im Hinblick individuelle Behandlungsstrategien, neue den Intelligenz.

---

Predictive Modeling of Acute Respiratory Distress Syndrome Using Machine Learning: Systematic Review and Meta-Analysis

Journal of Medical Internet Research, Journal Year: 2025, Volume and Issue: 27, P. e66615 - e66615

Published: May 13, 2025

Background Acute respiratory distress syndrome (ARDS) is a critical condition commonly encountered in the intensive care unit (ICU), characterized by high incidence and substantial mortality rate. Early detection accurate prediction of ARDS can significantly improve patient outcomes. While machine learning (ML) models are increasingly being used for prediction, there lack consensus on most effective model or methodology. This study first to systematically evaluate performance based multiple quantitative data sources. We compare effectiveness ML via meta-analysis, revealing factors affecting suggesting strategies enhance generalization accuracy. Objective aims existing through systematic review using metrics such as area under receiver operating characteristic curve, sensitivity, specificity, other relevant indicators. The findings will provide evidence-based insights support development more tools. Methods performed search across 6 electronic databases studies developing predictive ARDS, with cutoff date December 29, 2024. risk bias these was evaluated Prediction Risk Bias Assessment Tool. Meta-analyses investigations into heterogeneity were carried out Meta-DiSc software (version 1.4), developed Ramón y Cajal Hospital’s Clinical Biostatistics team Madrid, Spain. Furthermore, subgroup, meta-regression analyses explore sources comprehensively. Results achieved pooled curve 0.7407 ARDS. additional follows: sensitivity 0.67 (95% CI 0.66-0.67; P<.001; I²=97.1%), specificity 0.68 0.67-0.68; I²=98.5%), diagnostic odds ratio 6.26 4.93-7.94; I²=95.3%), positive likelihood 2.80 2.46-3.19; I²=97.3%), negative 0.51 0.46-0.57; I²=93.6%). Conclusions evaluates constructed various algorithms, results showing that demonstrates prediction. However, many still have limitations. During development, it essential focus quality, including reducing risk, designing appropriate sample sizes, conducting external validation, ensuring interpretability. Additionally, challenges physician trust need prospective validation must also be addressed. Future research should standardize optimize performance, how better integrate clinical practice diagnosis stratification. Trial Registration PROSPERO CRD42024529403; https://www.crd.york.ac.uk/PROSPERO/view/CRD42024529403

Language: Английский

---

Early Prediction of Mortality Risk in Acute Respiratory Distress Syndrome: Systematic Review and Meta-Analysis

Journal of Medical Internet Research, Journal Year: 2025, Volume and Issue: 27, P. e70537 - e70537

Published: May 20, 2025

Background Acute respiratory distress syndrome (ARDS) is a life-threatening condition associated with high mortality rates. Despite advancements in critical care, reliable early prediction methods for ARDS-related remain elusive. Accurate risk assessment crucial timely intervention and improved patient outcomes. Machine learning (ML) techniques have emerged as promising tools patients ARDS, leveraging complex clinical datasets to identify key prognostic factors. However, the efficacy of ML-based models remains uncertain. This systematic review aims assess value ML ARDS provide evidence supporting development simplified, clinically applicable scoring prognosis. Objective study systematically reviewed available literature on models, primarily aiming evaluate predictive performance these compare their conventional systems. It also sought limitations insights improving future tools. Methods A comprehensive search was conducted across PubMed, Web Science, Cochrane Library, Embase, covering publications from inception April 27, 2024. Studies developing or validating predicting were considered inclusion. The methodological quality bias assessed using Prediction Model Risk Bias Assessment Tool (PROBAST). Subgroup analyses performed explore heterogeneity model based dataset characteristics validation approaches. Results In total, 21 studies involving total 31,291 included. meta-analysis demonstrated that achieved relatively performance. training datasets, pooled concordance index (C-index) 0.84 (95% CI 0.81-0.86), while in-hospital prediction, C-index 0.83 0.81-0.86). external 0.81 0.78-0.84), corresponding 0.80 0.77-0.84). outperformed traditional tools, which lower area under receiver operating characteristic curve (ROC-AUC) standard systems 0.7 0.67-0.72). Specifically, 2 widely used systems, Sequential Organ Failure (SOFA) Simplified Physiology Score II (SAPS-II), ROC-AUCs 0.64 0.62-0.67) 0.70 0.66-0.74), respectively. Conclusions exhibited superior accuracy over suggesting potential use assessment. further research needed refine improve interpretability, enhance applicability. Future efforts should focus efficient, user-friendly integrate seamlessly into workflows. Such may facilitate identification high-risk patients, enabling interventions personalized, risk-based prevention strategies

Language: Английский

---

Novel machine learning models for the prediction of acute respiratory distress syndrome after liver transplantation

Frontiers in Artificial Intelligence, Journal Year: 2025, Volume and Issue: 8

Published: May 27, 2025

Early prediction of acute respiratory distress syndrome (ARDS) after liver transplantation (LT) facilitates timely intervention. We aimed to develop a predictor post-LT ARDS using machine learning (ML) methods. Data from 755 patients in the internal validation set and 115 external were retrospectively reviewed, covering demographics, etiology, medical history, laboratory results, perioperative data. According area under receiver operating characteristic curve (AUROC), accuracy, specificity, sensitivity, F1-value, performance seven ML models, including logistic regression (LR), decision tree, random forest (RF), gradient boosting tree (GBDT), naïve bayes (NB), light (LGBM) extreme (XGB) evaluated compared with lung injury scores (LIPS). 234 (30.99%) diagnosed. The RF model had best performance, an AUROC 0.766 (accuracy: 0.722, sensitivity: 0.617) comparable 0.844 0.809, 0.750) set. all models was better than LIPS (AUROC 0.692, 0.776). variables included age recipient, BMI, MELD score, total bilirubin, prothrombin time, operation standard urine volume, intake red blood cell infusion volume. firstly developed risk based on ameliorate clinical practice.

Language: Английский

---

Advancements and trends in nanomaterial development for acute respiratory distress syndrome

Asian Journal of Surgery, Journal Year: 2024, Volume and Issue: unknown

Published: Sept. 1, 2024

Language: Английский

---

Predictive Modeling of Acute Respiratory Distress Syndrome Using Machine Learning: Systematic Review and Meta-Analysis (Preprint)

Published: Sept. 18, 2024

Acute respiratory distress syndrome (ARDS) is a critical condition commonly encountered in the intensive care unit (ICU), characterized by high incidence and substantial mortality rate. Early detection accurate prediction of ARDS can significantly improve patient outcomes. While machine learning (ML) models are increasingly being used for prediction, there lack consensus on most effective model or methodology. This study first to systematically evaluate performance based multiple quantitative data sources. We compare effectiveness ML via meta-analysis, revealing factors affecting suggesting strategies enhance generalization accuracy. aims existing through systematic review using metrics such as area under receiver operating characteristic curve, sensitivity, specificity, other relevant indicators. The findings will provide evidence-based insights support development more tools. performed search across 6 electronic databases studies developing predictive ARDS, with cutoff date December 29, 2024. risk bias these was evaluated Prediction Risk Bias Assessment Tool. Meta-analyses investigations into heterogeneity were carried out Meta-DiSc software (version 1.4), developed Ramón y Cajal Hospital’s Clinical Biostatistics team Madrid, Spain. Furthermore, subgroup, meta-regression analyses explore sources comprehensively. achieved pooled curve 0.7407 ARDS. additional follows: sensitivity 0.67 (95% CI 0.66-0.67; <i>P</i>&lt;.001; <i>I</i>²=97.1%), specificity 0.68 0.67-0.68; <i>I</i>²=98.5%), diagnostic odds ratio 6.26 4.93-7.94; <i>I</i>²=95.3%), positive likelihood 2.80 2.46-3.19; <i>I</i>²=97.3%), negative 0.51 0.46-0.57; <i>I</i>²=93.6%). evaluates constructed various algorithms, results showing that demonstrates prediction. However, many still have limitations. During development, it essential focus quality, including reducing risk, designing appropriate sample sizes, conducting external validation, ensuring interpretability. Additionally, challenges physician trust need prospective validation must also be addressed. Future research should standardize optimize performance, how better integrate clinical practice diagnosis stratification. PROSPERO CRD42024529403; https://www.crd.york.ac.uk/PROSPERO/view/CRD42024529403

Language: Английский

---