Critical Care Medicine最新文献

Objectives: Artificial intelligence (AI) and machine learning (ML) are emerging as transformative tools in intensive care medicine. Nevertheless, despite the development of numerous AI/ML models, their integration into routine ICU practice remains limited. This concise review examines the role of AI and data science in critical care, with a focus on their contributions to safety and quality assurance, clinical processes improvements, and ICU management. By synthesizing current evidence, this review aims to highlight the opportunities and challenges associated with implementing AI-driven solutions in critical care settings.

Data sources: English-language articles were identified in PubMed using keywords related to AI, ML, ICU management, clinical decision support, and predictive analytics.

Study selection: Original research articles, reviews, letters, and commentaries relevant to AI/ML applications in ICU quality and performance assessment were included.

Data extraction: Relevant literature was identified, key findings were synthesized into a structured narrative review.

Data synthesis: The integration of AI and ML into ICU management leverages vast clinical data to evaluate ICU performance, measure risk factors, optimize workflows, and predict adverse events. ML-driven models can improve clinical decision-making and ICU management. Despite the promising results, real-world implementation requires rigorous validation and clinician adoption. AI-driven successful implementation in ICU comes with significant challenges.

Conclusions: AI and ML have the potential to transform ICU management. However, their success depends on validated methodologies, interoperable data frameworks, and interpretable models that clinicians can trust. Advancing AI use in the ICU demands a multidisciplinary effort to create adaptive, transparent, and clinically meaningful solutions that enhance patient care and improve workflow, while ensuring safety and efficiency.

Objectives: Delirium is common and harmful in the ICU. The Intensive Care Delirium Screening Checklist (ICDSC) and Confusion Assessment Method for the ICU (CAM-ICU) are validated tools recommended for delirium identification. However, the accuracy of bedside nurse-documented delirium assessments in the ICU is inconsistent, limiting utility in clinical research. We sought to evaluate and optimize agreement between bedside nurse-documented and trained researcher delirium assessments.

Design, setting, and patients: Critically ill adults with acute respiratory failure or sepsis in ICUs in large academic hospitals in a southwestern Pennsylvania health system were assessed daily for delirium by bedside nurses (using the ICDSC) and trained researchers (using the CAM-ICU). Using matched nurse-to-researcher delirium assessments, we categorized delirium status using validated cutoffs and evaluated agreement using Cohen's kappa. We derived and compared logistic regression models that used ICDSC documentation, mechanical ventilation status, and admission Sequential Organ Failure Assessment to predict delirium in noncomatose patients, using researcher CAM-ICU assessments as the reference standard. We internally validated models using ten-fold cross-validation.

Interventions: None.

Measurements and main results: From a sample of 1535 matched assessments of 279 patients, there was moderate agreement between bedside nurse assessments using the established ICDSC delirium/normal cutoff (ICDSC ≥ 4) and trained researcher assessments using the CAM-ICU (Cohen's kappa = 0.42). A logistic regression model informed by individual ICDSC components and clinical data predicted a positive research CAM-ICU with good discrimination (area under the curve = 0.87) and performed well in cross-validation (F1 score = 0.72). In sensitivity analyses, models with more limited ICDSC information demonstrated fair to good discriminatory ability (F1 = 0.60-0.70), with the validated cutoff model having the lowest performance.

Conclusions: A delirium model informed by bedside nurse ICDSC findings and clinical variables improves accuracy of delirium detected in the ICU and can be used in future pragmatic research that leverages large clinical datasets to advance understanding of delirium mechanisms, trajectories, and outcomes.

Objective: Patients with sepsis are prone to hypovolemia which can lead to hyperosmolar dehydration and result in intracellular volume depletion. This study aimed to assess the effect of hyperosmolar dehydration on the clinical outcomes of patients with sepsis and its potential as an indicator of optimal initial fluid management.

Design: A nationwide propensity score-matched cohort study analyzing data prospectively collected between September 2019 and December 2021.

Setting: Twenty tertiary- or university-affiliated hospitals in South Korea.

Patients: Adult patients with sepsis or septic shock admitted to the ICU.

Interventions: None.

Measurements and main results: Hyperosmolar dehydration was defined as serum osmolarity greater than or equal to 295 mOsm/L. The primary outcome, 30-day mortality, was compared using logistic regression adjusted for key prognostic factors in a 1:1 propensity score-matched cohort. Restricted cubic-spline models were used to analyze the clinical outcomes using the pre-ICU fluid volume as a continuous variable. Of the 4,487 patients, 2,605 (58.1%) had hyperosmolar dehydration. After matching, 1,537 pairs were analyzed. The 30-day mortality was higher in the hyperosmolar dehydration group (29.9%) than in the non-dehydration group (27.3%) (adjusted odds ratio, 1.18; 95% CI, 1.00-1.39). Liberal fluid management (greater than30 mL/kg) before ICU admission was associated with improved lactate levels in the hyperosmolar dehydration group ( p = 0.009) without increasing sequential organ failure assessment score ( p = 0.111). Among patients without dehydration, liberal fluid management was associated with an increased Sequential Organ Failure Assessment score ( p = 0.034) and a higher risk for mechanical ventilation ( p < 0.001), and without improving lactate levels ( p = 0.388).

Conclusions: Hyperosmolar dehydration at the diagnosis of sepsis was associated with increased 30-day mortality. A liberal fluid management benefits patients with hyperosmolar dehydration by improving lactate levels without increasing sequential organ failure assessment score. These findings highlight the importance of individualized fluid management based on the dehydration status in sepsis management.