AD-GCN: A novel graph convolutional network integrating multi-omics data for enhanced Alzheimer’s disease diagnosis
Open Access
- 2 June 2025
- journal article
- research article
- Published by Public Library of Science (PLoS) in PLOS ONE
- Vol. 20 (6) , e0325050
Abstract
Alzheimer’s disease (AD) etiology is complex, influenced by demographic risk factors such as age, sex, and educational level, alongside multi-omics factors derived from genomics, transcriptomics, and epigenomics. Advancements in multi-omics technology present both challenges and opportunities for AD diagnosis, enabling a more comprehensive understanding of the complex interactions among contributing factors, with the goal of improving diagnostic accuracy. To address this challenge, we propose a novel feature fusion approach in this study, AD-GCN, which integrates multi-omics data and their interaction networks to achieve more precise diagnosis and analysis of AD. In this study, we applied polygenic risk score and random forest algorithms for feature selection on genetic variation and methylation data. We then developed an AD-GCN for both multi-omics and single-omics classification tasks and compared its performance with that of machine learning ensemble methods. The experimental results demonstrated that multi-omics classification significantly outperformed single-omics classification, with AD-GCN surpassing the machine-learning ensembles. These findings highlight AD-GCN’s strong potential to enhance AD diagnosis and improve accuracy in differentiating disease stages by integrating interactions across omics data, laying a solid foundation for the development of more precise and personalized AD diagnostic models.Keywords
Funding Information
- the national key Research and Development Program of China (2022YFC2603305)
- Key Research and Development Program of Yunnan (202103AQ100002)
- Central Funds Guiding the Local Science and Technology Development (202207AB110017)
- Yunnan Academician and Expert Workstation (202205 AF150023)
- Yunnan Fundamental Research Projects (202201AY070001-192)
- the Scientific and Technological Innovation Team in Kunming Medical University (CXTD202215)
- Yunnan Provincial Science and Technology Program (202302AF080006)
This publication has 43 references indexed in Scilit:
- The association of GNB5 with Alzheimer disease revealed by genomic analysis restricted to variants impacting gene functionThe American Journal of Human Genetics, 2024
- APOE and Alzheimer disease risk across age, sex, race, ethnicity, and ancestry: An overview from 68,756 individualsAlzheimer's & Dementia, 2023
- Random forest kernel for high-dimension low sample size classificationStatistics and Computing, 2023
- Construction, evaluation, and AOP framework-based application of the EpPRS as a genetic surrogate for assessing environmental pollutantsEnvironment International, 2023
- Deep belief network-based approach for detecting Alzheimer's disease using the multi-omics dataComputational and Structural Biotechnology Journal, 2023
- Multimodal ensemble model for Alzheimer's disease conversion prediction from Early Mild Cognitive Impairment subjectsComputers in Biology and Medicine, 2022
- A Modified Borderline Smote with Noise Reduction in Imbalanced DatasetsWireless Personal Communications, 2021
- Using machine learning approaches for multi-omics data analysis: A reviewBiotechnology Advances, 2021
- A tutorial on conducting genome‐wide association studies: Quality control and statistical analysisInternational Journal of Methods in Psychiatric Research, 2018
- Modeling Linkage Disequilibrium Increases Accuracy of Polygenic Risk ScoresThe American Journal of Human Genetics, 2015