作者: Stephen Hogan , Andrew Gammie , Paul Abrams
DOI: 10.1002/NAU.22209
关键词: Natural language processing 、 Sample (statistics) 、 Event (probability theory) 、 Medicine 、 Urinary tract disorder 、 Artificial intelligence
摘要: Aims The aims of this study are to compile a list common features and artefacts found in urodynamics, produce definitions these features, describe any necessary remedial action. An image word description for each event has been included as well statistics providing information on the prevalence frequency event. Methods In order identify most 200 consecutive urodynamic traces were reviewed. A random 10% sample was cross-checked ensure classification accuracy. To extract significant pressure peaks from data, an algorithm written capable detecting initial resting updating it necessary. Significant defined those that differed values by 10 cmH2O or more. When describing events, standard sources consulted published definitions. The images selected typical examples but do not represent variation can occur between examples. patients whose files used suffered variety lower urinary tract disorders so is likely they cover all important events. Results In total 10,355 events identified classified into 19 different categories. For category, description, example action included. Where exist, new ones proposed. Conclusions All have descriptions one article first time. Neurourol. Urodynam. 31:1104–1117, 2012. © 2012 Wiley Periodicals, Inc.