Michele Introna

Target-controlled infusion (TCI) systems estimate plasma (Cp) and effect-site (Ce) drug concentrations using pharmacokinetic-pharmacodynamic (PK/PD) models rather than direct measurements. Because multiple models are used in clinical practice, predicted concentrations at the same clinical endpoint may differ substantially. Loss of consciousness (LOC) during anesthesia induction provides a clinically relevant reference point for examining these differences. We conducted a retrospective analysis of patients undergoing elective brain or spine surgery under propofol-based anesthesia. The propofol dose required to induce LOC was recorded, and plasma (Cp) and effect-site (Ce) concentrations at LOC were retrospectively estimated using four widely used PK/PD models (Marsh, Modified Marsh, Schnider, and Eleveld) through TCI simulation. Differences in model-derived concentrations were analyzed using mixed two-way ANOVA, with Model as a within-subject factor and Surgical Group (brain vs spine surgery) as a between-subject factor. When significant interactions were observed, post hoc comparisons were performed on log-transformed Ce values. Model-derived concentrations at LOC differed markedly across PK/PD models. Mean Ce values ranged from approximately 1.5 µg/mL with the Marsh and Eleveld models to 4.6 µg/mL with the Schnider model. Mixed ANOVA showed a strong model effect on both Ce and Cp (p < 0.001). A significant Model × Surgical Group interaction was observed for Ce but not for Cp. Patients undergoing brain surgery received lower propofol doses and reached LOC at lower model-predicted effect-site concentrations than patients undergoing spine surgery. Predicted propofol concentrations at LOC vary considerably across PK/PD models despite identical clinical dosing, highlighting the model-dependent nature of TCI targets. These findings underscore the importance of explicit model specification and adequate user understanding when interpreting TCI-derived concentrations in both clinical practice and research.

Recent correspondence in the Journal of Clinical Monitoring and Computing has raised concerns about the transparency of implementing the Eleveld pharmacokinetic/pharmacodynamic model for propofol in commercial target-controlled infusion systems. While these models are scientifically robust, limited insight into software-level implementation may hinder independent verification and clinical confidence. Addressing this requires not only improved transparency and validation but also adequate clinician training, particularly as intravenous anesthesia techniques gain prominence in response to evolving regulatory and environmental pressures.

Artificial intelligence (AI) has shown considerable potential in perioperative monitoring, particularly in its application to electroencephalogram (EEG) analysis for assessing the depth of anesthesia. AI methods may enable the dynamic recognition of complex time-frequency EEG patterns and the adaptation of monitoring strategies to patient-specific brain responses. Convolutional neural networks, artificial neural networks, and hybrid deep learning models have reported encouraging results in detecting anesthetic states, estimating bispectral index values, and identifying relevant EEG features - such as alpha-delta shifts or burst suppression - without relying on manual feature engineering. Parallel efforts using virtual and augmented reality platforms suggest possible benefits for anesthesiologist training in EEG interpretation and pharmacologic titration. Despite these advances, important limitations constrain clinical translation. A major challenge is the absence of standardized EEG pattern definitions across anesthetic agents and patient groups, limiting model generalizability. Restricted interoperability between EEG monitors and electronic health records, coupled with proprietary data formats, reduces access to raw EEG signals and hampers large-scale development. Privacy and governance requirements add further barriers to data integration. Methodologically, many studies are affected by insufficient internal validation, suboptimal reporting, and testing in experimental rather than real-world conditions, reducing their translational value. While AI could eventually improve anesthetic precision and safety through EEG-guided approaches, realizing this potential will require transparent algorithms, multicenter and heterogeneous datasets, and robust interoperability and data-sharing standards. Only through such coordinated efforts can these tools evolve from promising research applications into reliable components of routine anesthetic care.

Target-controlled infusion (TCI) systems, originally developed for intravenous drug administration of anesthetic drugs, enable precise drug delivery based on pharmacokinetic-pharmacodynamic (PKPD) models. While widely used in the operating room, their application in the intensive care unit (ICU) remains limited despite the complexity of drug dosing in critically ill patients. This scoping review evaluates existing evidence on the use of TCI systems in ICU settings, focusing on sedation, analgesia, and antibiotic administration. A systematic literature search was conducted in Medline and the Cochrane Library up to August 2024. Studies involving adult ICU patients receiving TCI for any drug were included. Data were extracted on study design, PKPD models, performance metrics, and clinical endpoints. A total of 281 studies were identified; 14 focused on sedatives, particularly propofol and remifentanil, using the Marsh and Minto models. TCI was associated with more stable sedation, improved titration, and fewer adverse events. Evidence on dexmedetomidine and midazolam was limited. Only three studies explored antibiotic TCI, mainly for vancomycin and cefepime, showing potential but lacking robust clinical validation. In conclusion, while current findings are promising, the evidence base supporting TCI—especially in the context of newer, generalized pharmacokinetic models and antibiotics—remains insufficient, and its broader application requires further clinical research and model validation.