天津医药

天津医药 ›› 2022, Vol. 50 ›› Issue (6): 648-652.doi: 10.11958/20211775

• 应用研究 • 上一篇    下一篇

基于脑电小波指数的人工智能给药系统的临床应用

何士凤，朱泽飞，张婉月，杨贯宇，郑红雨，孙振涛△   

  1. 郑州大学第一附属医院麻醉与围术期医学部（邮编450052） 
  • 收稿日期:2021-08-02 修回日期:2022-02-04 出版日期:2022-06-15 发布日期:2023-12-20
  • 通讯作者: 孙振涛 E-mail:gentlesun@126.com
  • 基金资助:
    河南省医学教育研究项目（Wjlx2019017）；河南省医学科技攻关计划项目（2018010006）；河南省卫生系统出国研修项目计划（2016021）

Clinical application of automated titration guided by EEG wavelet index

HE Shifeng, ZHU Zefei, ZHANG Wanyue, YANG Guanyu, ZHENG Hongyu， SUN Zhentao△   

  1. Department of Anesthesiology, Pain and Perioperative Medicine, the First Affiliated Hospital of Zhengzhou University, Zhengzhou 450052, China
  • Received:2021-08-02 Revised:2022-02-04 Published:2022-06-15 Online:2023-12-20
  • Contact: Zhen-Tao Sun E-mail:gentlesun@126.com

PDF (PC)

336

摘要： 摘要：目的　评估基于脑电小波指数的人工智能给药系统在临床中应用的可行性和安全性。方法　入选择期行腹腔镜结直肠癌根治术患者52例，以随机数字表法分为人工智能给药组（IT组）和手动调节组（CT组）。IT组在麻醉诱导和维持阶段均由人工智能给药系统基于脑电小波指数自动调节瑞芬太尼和丙泊酚输注速率。CT组在麻醉诱导和维持阶段均采用恒速泵来手动调节瑞芬太尼和丙泊酚输注速率。2组均设定目标镇静指数（WLi）、镇痛指数（PTi）为40～60。记录患者术中瑞芬太尼和丙泊酚用药剂量、干预调节的次数；记录患者诱导前（T0）、诱导后（T1）、手术开始即刻（T2）、手术开始后1 h（T3）、手术结束时（T4）的心率（HR）、平均动脉压（MAP）、血压差（ΔP）；记录术中给予血管活性药物的剂量、麻醉结束后拔管时间、术后麻醉恢复室（PACU）停留时间、不同血压水平持续时间占总手术时长的百分比、术中不良事件和术后7 d内的并发症。结果　与CT组相比，IT组术中丙泊酚的用量及干预调节次数明显降低（P＜0.05）；IT组术中低血压总占比及血管活性药物使用剂量明显低于CT组（P＜0.05）；2组术中不良事件发生率及术后7 d内并发症发生率差异无统计学意义（P＞0.05）。结论　基于脑电小波指数的人工智能给药系统可以减少术中丙泊酚的用量，降低术中低血压的发生率，减轻麻醉医师的工作负担，且不增加并发症的发生率，可安全用于腹腔镜结直肠癌根治术的患者。

关键词: 人工智能, 深度镇静, 镇痛, 低血压, 手术后并发症, 脑电小波指数

Abstract: Abstract： Objective　To evaluate the feasibility and safety of automated administration guided by electroencephalogram (EEG) wavelet index in clinical application. Methods　A total of 52 patients underwent laparoscopic colorectal cancer surgery were selected and divided into the artificial intelligence administration group (IT group) and the manual adjustment group (CT group) by random number table. In the IT group, the infusion rates of remifentanil and propofol were automatically adjusted by automated administration based on EEG wavelet index during anesthesia induction and maintenance. In the CT group, constant speed pumps were used to manually adjust the infusion rates of remifentanil and propofol during induction and maintenance of anesthesia. The target sedation index (WLi) and pain threshold index (PTi) were set at 40-60 in the both groups. Intraoperative doses of remifentanil and propofol and manual adjustment were recorded. Mean arterial pressure (MAP), blood pressure difference (ΔP) and heart rate (HR) were recorded before induction (T0), after induction (T1), immediately after surgery (T2), 1 h after surgery (T3) and at the end of surgery (T4). The intraoperative dose of vasoactive drugs, extubation time after anesthesia, the duration of postoperative anesthesia recovery room (PACU), the percentage of different blood pressure levels in the total operative duration and intraoperative adverse events and complications within 7 days after surgery were recorded. Results　Compared with the CT group, the amount of intraoperative propofol and manual adjustment was significantly decreased in the IT group (P＜0.05). There were no significant differences in the incidence of intraoperative adverse events and complications within 7 days after operation between the two groups (P＞0.05). Conclusion　The automated administration guided by EEG wavelet index can reduce the amount of intraoperative propofol, reduce the incidence of intraoperative hypotension and reduce the workload of anesthesiologists, without increasing the incidence of complications, which can be safely used in patients with laparoscopic radical colorectal cancer surgery.

Key words: artificial intelligence, deep sedation, analgesia, hypotension, postoperative complications, wavelet index