University Sétif 1 FERHAT ABBAS Faculty of Sciences
Détail de l'auteur
Auteur Ghozlane Hadri |
Documents disponibles écrits par cet auteur
Ajouter le résultat dans votre panier Affiner la recherche
Proteomic profiles selection and patients’ cancer classification based on bio-inspired algorithms and deep learning / Ghozlane Hadri
Titre : Proteomic profiles selection and patients’ cancer classification based on bio-inspired algorithms and deep learning Type de document : texte imprimé Auteurs : Ghozlane Hadri, Auteur ; Hassiba Boubadja, Auteur ; Abdelouahab Moussaoui, Directeur de thèse Editeur : Sétif:UFA1 Année de publication : 2023 Importance : 1 vol (127 f .) Format : 29cm Langues : Français (fre) Catégories : Thèses & Mémoires:Informatique Mots-clés : cancer diagnosis,
biologically inspired methodsIndex. décimale : 004 Informatique Résumé : Cancer research has made significant advancements in recent years. Utilizing high throughput technology and
advancements in artificial intelligence, it is now possible to enhance cancer diagnosis and targeted therapy
through the analysis of clinical and omics profiles. However, the abundance of available data, especially gene
expressing data, presents a formidable challenge due to its high dimensionality.
To address these challenges, we have developed two base line solutions the first is a supervised integrative
convolutional autoencoder(SICAE) that predicts clinical outpouts based on proteomic data. While the second
is a deep learning biologically inspired methods, specifically PSO-SICAE (Particle Swarm Optimization
- SICAE. The primary objective of this model is to identify biomarkers directly associated with the overall
survival of patients with Brain Lower Grade Glioma (LGG) cancer and the occurrence of new tumor events
after initial treatment in Stomach Adenocarcinoma (STAD), Pancreatic Adenocarcinoma (PAAD), Liver Hepatocellular
Carcinoma (LIHC), Cholangiocarcinoma (CHOL), and Glioblastoma Multiforme (GBM) cancers.
To thoroughly evaluate the performance of our proposed model, we conducted a comprehensive experimental
study involving both digestive system cancers and nervous system cancers.
In this study, we compared our model with traditional machine learning algorithms for feature selection, deep
learning models, and various biologically inspired algorithms, using numerous evaluation metrics. The results
obtained are highly promising, demonstrating the effectiveness of our proposed framework. Our model outperformed
other algorithms and models in terms of accuracy (76%), AUC (62%), and several other evaluation
metrics.Côte titre : MAI/0715 En ligne : https://drive.google.com/file/d/1ar-eFbVbLGL6ttBKmgvfoILoyaGnfk-K/view?usp=drive [...] Format de la ressource électronique : Proteomic profiles selection and patients’ cancer classification based on bio-inspired algorithms and deep learning [texte imprimé] / Ghozlane Hadri, Auteur ; Hassiba Boubadja, Auteur ; Abdelouahab Moussaoui, Directeur de thèse . - [S.l.] : Sétif:UFA1, 2023 . - 1 vol (127 f .) ; 29cm.
Langues : Français (fre)
Catégories : Thèses & Mémoires:Informatique Mots-clés : cancer diagnosis,
biologically inspired methodsIndex. décimale : 004 Informatique Résumé : Cancer research has made significant advancements in recent years. Utilizing high throughput technology and
advancements in artificial intelligence, it is now possible to enhance cancer diagnosis and targeted therapy
through the analysis of clinical and omics profiles. However, the abundance of available data, especially gene
expressing data, presents a formidable challenge due to its high dimensionality.
To address these challenges, we have developed two base line solutions the first is a supervised integrative
convolutional autoencoder(SICAE) that predicts clinical outpouts based on proteomic data. While the second
is a deep learning biologically inspired methods, specifically PSO-SICAE (Particle Swarm Optimization
- SICAE. The primary objective of this model is to identify biomarkers directly associated with the overall
survival of patients with Brain Lower Grade Glioma (LGG) cancer and the occurrence of new tumor events
after initial treatment in Stomach Adenocarcinoma (STAD), Pancreatic Adenocarcinoma (PAAD), Liver Hepatocellular
Carcinoma (LIHC), Cholangiocarcinoma (CHOL), and Glioblastoma Multiforme (GBM) cancers.
To thoroughly evaluate the performance of our proposed model, we conducted a comprehensive experimental
study involving both digestive system cancers and nervous system cancers.
In this study, we compared our model with traditional machine learning algorithms for feature selection, deep
learning models, and various biologically inspired algorithms, using numerous evaluation metrics. The results
obtained are highly promising, demonstrating the effectiveness of our proposed framework. Our model outperformed
other algorithms and models in terms of accuracy (76%), AUC (62%), and several other evaluation
metrics.Côte titre : MAI/0715 En ligne : https://drive.google.com/file/d/1ar-eFbVbLGL6ttBKmgvfoILoyaGnfk-K/view?usp=drive [...] Format de la ressource électronique : Exemplaires (1)
Code-barres Cote Support Localisation Section Disponibilité MAI/0715 MAI/0715 Mémoire Bibliothéque des sciences Anglais Disponible
Disponible