Ben-Israel Nir, Kliger Mark, Zuckerman Galit, Katz Yeshayahu, Edry Ruth
Medasense Biometrics Ltd., Ramat Yishai, Israel.
J Clin Monit Comput. 2013 Dec;27(6):659-68. doi: 10.1007/s10877-013-9487-9. Epub 2013 Jul 9.
The aim of the present study was to develop and validate an objective index for nociception level (NoL) of patients under general anesthesia, based on a combination of multiple physiological parameters. Twenty-five patients scheduled for elective surgery were enrolled. For clinical reference of NoL, the combined index of stimulus and analgesia was defined as a composite of the surgical stimulus level and a scaled effect-site concentration of opioid. The physiological parameters heart rate, heart rate variability (0.15-0.4 Hz band power), plethysmograph wave amplitude, skin conductance level, number of skin conductance fluctuations, and their time derivatives, were extracted. Two techniques to incorporate these parameters into a single index representing the NoL have been proposed: NoLlinear, based on an ordinary linear regression, and NoLnon-linear, based on a non-linear Random Forest regression. NoLlinear and NoLnon-linear significantly increased after moderate to severe noxious stimuli (Wilcoxon rank test, p < 0.01), while the individual parameters only partially responded. Receiver operating curve analysis showed that NoL index based on both techniques better discriminated noxious and non-noxious surgical events [area under curve (AUC) = 0.97] compared with individual parameters (AUC = 0.56-0.74). NoLnon-linear better ranked the level of nociception compared with NoLlinear (R = 0.88 vs. 0.77, p < 0.01). These results demonstrate the superiority of multi-parametric approach over any individual parameter in the evaluation of nociceptive response. In addition, advanced non-linear technique may have an advantage over ordinary linear regression for computing NoL index. Further research will define the usability of the NoL index as a clinical tool to assess the level of nociception during general anesthesia.
本研究的目的是基于多个生理参数的组合，开发并验证一种用于全身麻醉患者痛觉水平（NoL）的客观指标。招募了25例计划接受择期手术的患者。作为NoL的临床参考，刺激与镇痛的综合指标被定义为手术刺激水平与阿片类药物效应室浓度的标度值的复合指标。提取了生理参数心率、心率变异性（0.15 - 0.4Hz频段功率）、体积描记波波幅、皮肤电导水平、皮肤电导波动次数及其时间导数。提出了两种将这些参数整合为一个代表NoL的单一指标的技术：基于普通线性回归的NoLlinear和基于非线性随机森林回归的NoLnon-linear。中度至重度有害刺激后，NoLlinear和NoLnon-linear显著增加（Wilcoxon秩和检验，p < 0.01），而单个参数仅部分有反应。受试者工作特征曲线分析表明，与单个参数（AUC = 0.56 - 0.74）相比，基于这两种技术的NoL指标能更好地区分有害和无害手术事件[曲线下面积（AUC）= 0.97]。与NoLlinear相比，NoLnon-linear对痛觉水平的排序更好（R = 0.88对0.77，p < 0.01）。这些结果证明了在评估伤害性反应方面多参数方法优于任何单个参数。此外，先进的非线性技术在计算NoL指标方面可能比普通线性回归具有优势。进一步的研究将确定NoL指标作为评估全身麻醉期间痛觉水平的临床工具的可用性。
