Skip to Main content Skip to Navigation

Nouvelles approches pour la détection de relations utiles dans les processus : application aux parcours de santé

Abstract : Ever since the post-World War II baby-boom, France has had to cope with an aging population and with pathologies that have become chronic. These new health problems imply more reccurent, more complex and multidisciplinary medical care. However, multiple obstacles related to the evolution of the society or to the internal organization of the healthcare system hinder the development of new and more adapted health procedures to meet new medical needs. In a context where health expanses have to be reduced, it is necessary to have a better management of health processes.Our work aims at proposing a set of methods to gain an understanding of the elderly health path in the Auvergne region. To this end, a description of the elderly health path is necessary in order to have this missing overview. Moreover, this will enable the identification of the stakeholders, their interactions and the constraints to which they are submitted in the different health procedures.The work presented in this thesis focuses on two problems. The first one consists in developing techniques to efficiently model health paths. With these models, we will be able to analyze how the different segments of a medical care are ordered. The second problem consists in developing techniques to extract relevant information from the data, according to a predefined business point of view, specific to the user who analyzes the model. This knowledge will enable the detection of frequent or abnormal parts of a health path.To resolve these problems, the methods we propose in this thesis are related to process mining and data mining. These disciplines aim at exploiting data available in today's information systems in order to discover useful knowledge. In a first part, we propose a methodology that relies on the expressive power of process models to extract relevant information. In a second part, we propose techniques to build local process models that represent interesting fragments of behavior.The experiments we performed show the efficiency of the methods we propose. Moreover, we analyze data from different application domains to prove the genericity of the developed techniques.
Document type :
Complete list of metadata

Cited literature [165 references]  Display  Hide  Download
Contributor : Abes Star :  Contact
Submitted on : Tuesday, November 27, 2018 - 8:43:09 AM
Last modification on : Wednesday, February 24, 2021 - 4:24:02 PM
Long-term archiving on: : Thursday, February 28, 2019 - 12:48:52 PM


Version validated by the jury (STAR)


  • HAL Id : tel-01935705, version 1



Benjamin Dalmas. Nouvelles approches pour la détection de relations utiles dans les processus : application aux parcours de santé. Autre [cs.OH]. Université Clermont Auvergne, 2018. Français. ⟨NNT : 2018CLFAC005⟩. ⟨tel-01935705⟩



Record views


Files downloads