DI-UMONS : Dépôt institutionnel de l’université de Mons

Recherche transversale
(titres de publication, de périodique et noms de colloque inclus)
2018-07-19 - Colloque/Présentation - poster - Anglais - 0 page(s)

Itani Sarah , Rossignol Mandy , Lecron Fabian , Fortemps Philippe , "On the Involvement of the Limbic System in the Diagnosis of Attention Deficit Hyperactivity Disorder" in International Neuropsychological Society 2018 Mid-Year Meeting, Prague, République tchèque, 2018

  • Codes CREF : Intelligence artificielle (DI1180), Modèles mathématiques d'aide à la décision (DI1151), Technologies de l'information et de la communication (TIC) (DI4730), Neurosciences cognitives (DI4296), Informatique mathématique (DI1160)
  • Unités de recherche UMONS : Psychologie cognitive et Neuropsychologie (P325), Management de l'Innovation Technologique (F113), Mathématique et Recherche opérationnelle (F151)
  • Instituts UMONS : Institut de Recherche en Technologies de l’Information et Sciences de l’Informatique (InforTech), Institut des Sciences et Technologies de la Santé (Santé)
  • Centres UMONS : Centre de recherche interdisciplinaire en Psychophysiologie et Electrophysiologie de la cognition (CIPsE)

Abstract(s) :

(Anglais) Objective: the present work aimed at developing a predictive model for Attention Deficit Hyperactivity Disorder (ADHD) diagnosis aid, based on a sample of typically developing and ADHD patients. Indeed, the advent of machine learning has given translational neuroscience research new insights, in particular as regards the issue of diagnosis aid. Yet ADHD is among neuropathologies that still need to be addressed through a more objective diagnosis. Methods: we considered a sample of patients extracted from the ADHD-200 collection. We used decision trees as predictive models for their readability. Indeed, these models propose a sequence of questions on a patient to predict his/her medical condition. However, this readability does not necessarily ensure that the model provides consistent explanations. The study was thus achieved under a bottom-up methodology, conducted in a semi-knowledge guided framework. First, we lead a dimensionality reduction procedure on the dataset of phenotypic and resting-state functional magnetic resonance imaging features. This method raised the most significant features; the latter were then used to develop a decision tree which was interpreted to extract explanations on ADHD. This knowledge was used in the second part of the study, in selecting explicitly significant diagnosis elements (brain zones, phenotypic features) based on which a final decision tree was developed. Results: our results suggest the involvement of the limbic system in the diagnosis of ADHD. Besides, the final prediction accuracy is better than those of the recent literature. Conclusions: to sum up, it appears the interaction of machine learning and neuroscience is promising in the perspective of (1) developing models for diagnosis aid, (2) finding elements able to explain mental health issues like ADHD.