Patient monitors play an important role during surgery, especially in anesthetic management. They measure, process and display waveform signals and derive clinical variables from them. Artifacts -- despite filtering techniques designed to reduce them -- affect these signals and, in turn, the accuracy of the parameters derived from them. Bala Gopakumaran and colleagues from the Cleveland Clinic Foundation in Ohio have reviewed the problems posed by nonphysiological artifacts found commonly in ECG, blood pressure, noninvasive blood pressure, pulse oximetry, capnograpy and temperature data. For each of these signals, they compared artifact reduction methods used in patient monitors from four major manufacturers. Although they highlighted technologies that have enjoyed some success -- such as Raman scattering spectroscopy for capnography -- they noted that most monitors employ linear filters that fail to remove artifacts whose frequencies overlap with that of the desired signal. These same machines use averaging techniques to filter parametric data, which yield displayed measurements that do not reflect the most current patient status. The scientists also examined methods that the research community had proposed to reduce artifacts and found artificial intelligence techniques such as fuzzy logic, artificial neural networks and genetic algorithms particularly promising. Because patient monitors receive information from multiple biomedical signals, some of which result from the same physiological phenomenon, a system that incorporates such technology could -- as with a clinician -- automatically reference alternate waveforms to determine whether a suspicious data set contains artifacts. Signal processing methods such as Kalman filters and wavelet filtering could provide better data to the artificial intelligence systems and more effectively extract data of interest from an artifactual signal. The authors also outlined the functions that patient monitors should carry out and stressed the need to incorporate improved, reliable artifact correction methods into the next-generation models. (Anesthesia & Analgesia, November 2006, p. 1196).