A Deep Learning–Based Framework for Supporting Clinical Diagnosis of Glioblastoma Subtypes

Abstract

Understanding molecular features that facilitate aggressive phenotypes in glioblastoma multiforme (GBM) remains a major clinical challenge. Accurate diagnosis of GBM subtypes, namely classical, proneural, and mesenchymal, and identification of specific molecular features are crucial for clinicians for systematic treatment. We develop a biologically interpretable and highly efficient deep learning framework based on a convolutional neural network for subtype identification. The classifiers were generated from high-throughput data of different molecular levels, i.e., transcriptome and methylome. Furthermore, an integrated subsystem of transcriptome and methylome data was also used to build the biologically relevant model. Our results show that deep learning model outperforms the traditional machine learning algorithms. Furthermore, to evaluate the biological and clinical applicability of the classification, we performed weighted gene correlation network analysis, gene set enrichment, and survival analysis of the feature genes. We identified the genotype–phenotype relationship of GBM subtypes and the subtype-specific predictive biomarkers for potential diagnosis and treatment.

ecGBMsub: an integrative stacking ensemble model framework based on eccDNA molecular profiling for improving IDH wild-type glioblastoma molecular subtype classification

“…First, our model was validated in two large-scale independent cohorts, ensuring its generalization ability. Munquad et al developed a biologically interpretable deep learning framework based on a convolutional deep neural network (CDNN) for subtype classification ( Munquad et al, 2022 ). Their model integrated transcriptome and methylome to accurately predict molecular subtypes.…”

Immune landscape-based machine-learning–assisted subclassification, prognosis, and immunotherapy prediction for glioblastoma

“…In addition, there have been existing ML models for GBM subtype classification based on different data; for example, Munquad et al. ( 84 ) utilized transcriptome and methylome data to construct classifiers through several ML algorithms, and the best model presents an accuracy of 87.5% on the testing data and 94.48% on external data. Macyszyn et al.…”

Artificial Intelligence in battling infectious diseases: A transformative role

“…40 Another study focused on subtype-specific predictive biomarker discovery, applicable to disease diagnosis and treatment. 41 Scmap employs a graph-based clustering technique to assess the maximum similarity between cells in both reference and query data, enabling…”

“…To address the challenge of identifying biomarkers in every cell cluster, a framework based on regularized multitask learning was developed for simultaneous prediction of subpopulations related to specified cell types 40 . Another study focused on subtype‐specific predictive biomarker discovery, applicable to disease diagnosis and treatment 41 . Scmap employs a graph‐based clustering technique to assess the maximum similarity between cells in both reference and query data, enabling the identification of distinct clusters corresponding to different cell types 42 .…”