Skip to Main content Skip to Navigation

Étude d'algorithmes d'apprentissage artificiel pour la prédiction de la syncope chez l'homme

Abstract : Syncope is considered as a common pathology, although sometimes its cause cannot be clearly diagnosed. In this case and when syncope is frequently experienced, the patient can have a head-upright tilt test. This examination is based on the reproduction of symptoms of the syncope ; however, its major drawback is the duration of the examination, which can take up to one hour. Therefore, reducing the examination time would decrease its cost and improve the comfort of the patient. This is the challenge of this thesis, which tries to predict the appearance of the symptoms of the syncope before the end of the tilt test. During the research, two areas of study became important : data mining and development of models used to predict the tilt-test result. Both areas use algorithms coming from machine learning, enabling the acquisition and extraction of relevant knowledge on data sets. Published works give many methods, which have enabled the extraction of some pertinent characteristics. With these, robust and ecient models have been constructed, which have enabled the prediction of the tilt-test results in the rst ten minutes of the examination. Also, the performance has been improved by the development of new techniques of data mining, enabling more ecient analysis of data. These methods have been used for the selection of the variables and for the interpretation of the non-linear projection techniques. Even though these methods have been developed for this research, they have shown interesting performances during tests on other data sets.
Document type :
Complete list of metadatas
Contributor : Anne-Marie Plé <>
Submitted on : Thursday, March 18, 2010 - 4:01:47 PM
Last modification on : Monday, October 19, 2020 - 11:12:13 AM
Long-term archiving on: : Wednesday, November 30, 2016 - 3:48:42 PM


  • HAL Id : tel-00465008, version 1


Mathieu Feuilloy. Étude d'algorithmes d'apprentissage artificiel pour la prédiction de la syncope chez l'homme. Informatique [cs]. Université d'Angers, 2009. Français. ⟨tel-00465008⟩



Record views


Files downloads