Developmental Differences in Gamma-Band Brain Connectivity During Propofol and Sevoflurane Anesthesia in Children

Abstract Background Propofol and sevoflurane are commonly used pediatric anesthetics with distinct molecular targets; however, their differential modulation of developing brain networks across age groups remains poorly understood. Unlike scalp low-frequency EEG, source-space gamma-band functional connectivity reflects cortical processing and may offer a sensitive window into drug-specific effects. This study aimed to identify distinct gamma-band connectivity patterns distinguishing propofol from sevoflurane anesthesia using explainable machine learning. Methods We analyzed 19-lead EEG data from 196 children undergoing general anesthesia, divided into an internal derivation cohort (n = 150) and an external validation cohort (n = 46). Patients were stratified into three age groups (≤ 42 months(G1), 43–83 months(G2), and ≥ 84 months(G3)). Cortical activity was reconstructed via source imaging using eLORETA. Machine learning models were developed to classify anesthetic types based on gamma-band connectivity features. Model performance was evaluated using internal and external validation, and Shapley Additive Explanations (SHAP) were applied to identify key anatomical regions driving network reorganization. Results Model predictive performance exhibited a distinct age gradient. The internal validation Area Under the Curve (AUC) significantly improved from G1 (e.g., MLP AUC = 0.703), G2(MLP AUC = 0.794), to G3 (MLP AUC = 0.922), with external validation confirming robust generalization in older children (e.g., SVM AUC = 0.773). SHAP analysis based on 150 optimal features revealed age-dependent network shifts: distinguishing anesthetics relied on primary sensory-limbic interactions in G1, shifted to default mode network and orbitofrontal connectivity in G2, and depended on long-range parahippocampal and occipito-frontal networks in G3. Conclusions Source-space gamma -band connectivity combined with explainable machine learning effectively distinguishes propofol from sevoflurane anesthesia in children. These findings reveal that distinct, drug-specific modulation patterns mature alongside the developing brain, highlighting a pronounced age gradient that lays the groundwork for age-adapted precision anesthesia monitoring.

Authors

Institutions

Publication Details

Journal
Research Square
Published
2026-11-12
DOI
https://doi.org/10.21203/rs.3.rs-11430305/v1
Primary Topic
Anesthesia and Sedative Agents
Type
preprint
Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
NOV
preprint

Developmental Differences in Gamma-Band Brain Connectivity During Propofol and Sevoflurane Anesthesia in Children

Liming Cheng, Jingping Wang, Junpeng Zhang, Lifang Yang et al.
Research Square
Anesthesia and Sedative Agents
preprint

Developmental Differences in Gamma-Band Brain Connectivity During Propofol and Sevoflurane Anesthesia in Children

Liming Cheng, Jingping Wang, Junpeng Zhang, Lifang Yang, Rui Zhao
preprint en

Abstract

Abstract Background Propofol and sevoflurane are commonly used pediatric anesthetics with distinct molecular targets; however, their differential modulation of developing brain networks across age groups remains poorly understood. Unlike scalp low-frequency EEG, source-space gamma-band functional connectivity reflects cortical processing and may offer a sensitive window into drug-specific effects. This study aimed to identify distinct gamma-band connectivity patterns distinguishing propofol from sevoflurane anesthesia using explainable machine learning. Methods We analyzed 19-lead EEG data from 196 children undergoing general anesthesia, divided into an internal derivation cohort (n = 150) and an external validation cohort (n = 46). Patients were stratified into three age groups (≤ 42 months(G1), 43–83 months(G2), and ≥ 84 months(G3)). Cortical activity was reconstructed via source imaging using eLORETA. Machine learning models were developed to classify anesthetic types based on gamma-band connectivity features. Model performance was evaluated using internal and external validation, and Shapley Additive Explanations (SHAP) were applied to identify key anatomical regions driving network reorganization. Results Model predictive performance exhibited a distinct age gradient. The internal validation Area Under the Curve (AUC) significantly improved from G1 (e.g., MLP AUC = 0.703), G2(MLP AUC = 0.794), to G3 (MLP AUC = 0.922), with external validation confirming robust generalization in older children (e.g., SVM AUC = 0.773). SHAP analysis based on 150 optimal features revealed age-dependent network shifts: distinguishing anesthetics relied on primary sensory-limbic interactions in G1, shifted to default mode network and orbitofrontal connectivity in G2, and depended on long-range parahippocampal and occipito-frontal networks in G3. Conclusions Source-space gamma -band connectivity combined with explainable machine learning effectively distinguishes propofol from sevoflurane anesthesia in children. These findings reveal that distinct, drug-specific modulation patterns mature alongside the developing brain, highlighting a pronounced age gradient that lays the groundwork for age-adapted precision anesthesia monitoring.

Research Square
Harvard University (US), Sichuan University (CN), Kunming Children's Hospital (CN), Tongren Hospital (CN), Xi'an Jiaotong University (CN)
Good health and well-being
Anesthesia and Sedative Agents
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.

Laika access coming soon