Korean Journal of Anesthesiology最新文献

Aging is associated with widespread structural and functional changes in the brain including reduced neural plasticity, slower information processing, and impaired network integration. These age-related alterations influence the brain's response to anesthetic agents, particularly electroencephalography (EEG) activity. This narrative review summarizes the characteristic EEG features induced by commonly used hypnotic agents such as propofol, inhaled anesthetics, dexmedetomidine, ketamine, and remimazolam in elderly patients and examines how aging modulates these responses. With increasing age, EEG power shows a global decline, most prominently in the alpha frequency band (8-13 Hz), reflecting reduced thalamocortical and cortical activity. Peak alpha frequency slows progressively with age, and background EEG also often exhibits characteristic slowing, both of which are associated with cognitive decline. In addition, EEG reactivity to external stimuli diminishes, and integrative brain activity, representing coordinated processing across cortical regions, is reduced in older adults. Frontoparietal feedback connectivity, essential for conscious perception and information integration, is particularly weak in the elderly. These changes are further exacerbated under anesthesia, as general anesthetics disrupt top-down connectivity and reduce network integration. Graph-theoretical EEG analyses reveal age-related reductions in global efficiency, modularity, and small-world properties, which are signatures of a less efficient, more random, and fragmented brain network. Understanding these age-specific EEG alterations can improve intraoperative monitoring, anesthetic titration, and development of age-tailored EEG-guided strategies. Future research should aim to validate EEG biomarkers that reliably reflect anesthetic depth and brain health in elderly populations, thereby fostering safer anesthesia care in the aging population.

Background: Recent advancements in critical care have highlighted the need for comprehensive, multimodal datasets to support clinical decision-making and advancing artificial intelligence (AI) research. However, such datasets are scarce in Asia. We developed the Korean Multi-Institutional Multimodal Intensive Care (K-MIMIC) dataset by integrating structured electronic medical records (EMRs), high-resolution bio-signals, and medical imaging from multiple hospitals in Korea.

Methods: This retrospective multicenter study collected intensive care unit (ICU) data from 278,274 patients admitted to 71 ICUs across 10 hospitals between 2001 and 2023. The data modalities included structured EMRs, physiological waveforms, and imaging studies. Data extraction followed standardized protocols and de-identification procedures in compliance with the Korean Health Data Utilization Guidelines. Multimodal linkage was achieved at the patient level to enable temporal trajectory analysis.

Results: The K-MIMIC dataset contains 287,274 ICU admissions from 241,805 unique patients, including 22,588 bio-signal files and 496,999 imaging studies, primarily chest X-rays aligned with EMRs. Nearly 47% of ICU admissions originated in the emergency department (ED). Elderly patients (65-90 years old) constituted the largest age group. Fifteen thousand, five hundred forty-eight patients had EMR data linked with both bio-signals and imaging, enabling full multimodal analyses.

Conclusions: The K-MIMIC is the first large-scale, multicenter, multimodal ICU dataset in Asia to provide a robust resource for critical care research, including AI-based prediction, monitoring, and longitudinal outcome studies. The dataset demonstrates the feasibility of secure and standardized ICU data integration across diverse institutions.

Background: Opioids remain widely used for postoperative pain control after total knee arthroplasty (TKA); however, concerns about adverse effects and dependency drive interest in opioid-sparing alternatives. This study evaluated the efficacy and safety of opioid-sparing patient-controlled analgesia (PCA) after TKA.

Methods: In this prospective, randomized, double-blind, non-inferiority study, 98 patients undergoing TKA under spinal anesthesia received either opioid-based PCA (continuous infusion of 1200 μg fentanyl, n = 49) or opioid-sparing PCA (continuous infusion of 150 mg ketorolac tromethamine and 100 mg nefopam hydrochloride, n = 49). Both groups received patient-controlled boluses of 300 μg fentanyl. The primary endpoint was the visual analog scale (VAS) pain score at rest on postoperative day (POD) 1, assessed using a 1.5-point non-inferiority margin. Secondary endpoints included additional analgesics, mobility, postoperative pain at rest and during ambulation, and adverse effects on PODs 1 and 2.

Results: The mean VAS score at rest on POD 1 was 5.45 ± 2.48 in the opioid-based PCA group and 5.90 ± 2.31 in the opioid-sparing PCA group. The mean difference was 0.45 points (95% CI, -0.36 to 1.25), within the prespecified non-inferiority margin. Pain scores at each time point were non-inferior in the opioid-sparing group, whereas rescue analgesic requirements were significantly reduced on POD 2 (P = 0.006). Nausea and vomiting on POD 1 were more frequent with opioid-based group (34.7% vs. 12.2%, P = 0.009).

Conclusions: Opioid-sparing PCA with ketorolac and nefopam provides non-inferior analgesia to opioid-based PCA, while reducing opioid consumption and drug-related adverse effects after TKA.

Background: Conventional machine learning (ML) models for predicting surgical outcomes have limitations in generalizability We explored large language models (LLMs) as scalable alternatives to conventional ML models in predicting postoperative outcomes, including in-hospital 30-day mortality, intensive care unit (ICU) admission, and acute kidney injury (AKI).

Methods: This study utilized the Informative Surgical Patient for Innovative Research Environment (INSPIRE) dataset (n = 80,985) from South Korea for model development and internal validation, and the Medical Informatics Operating Room Vitals and Events Repository (MOVER) dataset (n = 6,165) from the United States for external validation. The study compared three different LLMs-Generative Pre-trained Transformer [GPT]-4o, Llama-3-70B, and OpenBioLLM-70B-against MLs using various prompt engineering approaches. LLMs were evaluated with different model parameter quantizations (4-bit normalized floating point vs. 16-bit brain floating point).

Results: OpenBioLLM-70B were comparable to eXtreme Gradient Boosting (XGBoost) across all tasks (in-hospital 30-day mortality: area under receiver operating characteristic curve [AUROC] 0.782 [95% CI: 0.748-0.813] vs. 0.791 [95% CI: 0.753-0.825]; ICU admission: AUROC 0.595 [95% CI: 0.581-0.609] vs. 0.594 [95% CI: 0.580-0.608]; AKI: AUROC 0.830 [95% CI: 0.802-0.855] vs. 0.823 [95% CI: 0.792-0.851]) during external validation. Open-source LLMs maintained performance with 4-bit quantization, reducing computational requirements by 75%.

Conclusions: The findings support the versatility and efficiency of LLMs for clinical decision support through on-premises compatibility, addressing data privacy. Further validation with diverse datasets is needed to ensure their reliability and applicability across different perioperative settings.

Processed electroencephalogram (EEG) indices, such as the Bispectral Index, have markedly influenced anesthesia practice as they translate brain activity into simple numerical indices. Nevertheless, as the manufacturing algorithms are not disclosed, the underlying neurophysiology remains obscured. Additionally, these indices are often affected by electromyographic contamination, pharmacological variability, and patient-specific EEG heterogeneity. In contrast, an EEG spectrogram, or density spectral array, preserves the frequency- and time-resolved structures of cortical oscillations. This information is presented in a form that is both physiologically meaningful and clinically interpretable. In this review, we trace the evolution of anesthesia from an index-based to a spectrogram-guided approach, and summarize the clinical rationale for adopting the latter. Key applications of this approach include the use of frontal alpha power as a biomarker of cortical stability and postoperative brain health, the identification of nociceptive arousal through alpha dropout and beta or delta arousal patterns, and individualized titration of multimodal or age-specific anesthetic management. Although current devices lack standardized quantitative alpha metrics and have limited sensitivity for low-frequency brain wave components, structured EEG education programs have proven to be effective in terms of fostering spectrogram literacy among anesthesiologists. By combining neurophysiological precision with bedside practicality, the EEG spectrogram represents a pivotal advance toward individualized, mechanism-based, and brain-protective anesthesia, transforming anesthetic monitoring from mere algorithmic abstraction to cortical insight.