Urban R J, Johnson M L, Veldhuis J D
Department of Internal Medicine, University of Virginia School of Medicine, Charlottesville 22908.
Am J Physiol. 1989 Jul;257(1 Pt 1):E88-94. doi: 10.1152/ajpendo.1989.257.1.E88.
To investigate the nature of false-negative and false-positive errors in endocrine signal detection, we used a multiparameter convolution procedure to create random hormone secretory bursts. We observed that 1) data containing high signal frequency, amplitude, and/or duration can be analyzed at less stringent peak-detection thresholds; 2) for any given secretory burst amplitude, increasing the peak-detection threshold enhances positive accuracy but decreases sensitivity; 3) increased sampling frequency improves sensitivity but requires more stringent peak-detection thresholds; and 4) increasing noise diminishes both sensitivity and positive accuracy. We conclude that secretory properties, peak-detector thresholds, investigator-specified sampling intensity, and experimental variance all significantly influence false-positive and false-negative errors associated with the enumeration of episodic endocrine pulse signals. The present observations should offer objective principles to aid in the rational design and analysis of neuroendocrine studies of pulsatile physiological phenomena.
为了研究内分泌信号检测中假阴性和假阳性错误的本质，我们使用了一种多参数卷积程序来创建随机激素分泌脉冲。我们观察到：1）包含高信号频率、幅度和/或持续时间的数据可以在不太严格的峰值检测阈值下进行分析；2）对于任何给定的分泌脉冲幅度，提高峰值检测阈值可提高阳性准确率但降低灵敏度；3）增加采样频率可提高灵敏度，但需要更严格的峰值检测阈值；4）增加噪声会降低灵敏度和阳性准确率。我们得出结论，分泌特性、峰值检测器阈值、研究者指定的采样强度和实验方差均会显著影响与间歇性内分泌脉冲信号计数相关的假阳性和假阴性错误。本研究结果应能提供客观原则，以帮助合理设计和分析搏动性生理现象的神经内分泌研究。
