Quantum-Integrated Deep Learning Framework for Large-Scale Gene Expression Analysis and Predictive Modeling of Parkinson’s and Alzheimer’s disease

Abstract

Parkinson disease (PD) and Alzheimer disease (AD) are progressive neurodegenerative diseases that are marked by complicated molecular changes and shared pathogenesis. Due to its significant role in the appearance and further evolution of disease, transcriptomic deregulation is quite difficult to measure; the key issue is finding solid molecular signatures among hundreds of gene activity profiles. The objectives of the current research were to isolate differentially expressed genes (DEGs) that are related to PD and AD, as well as to determine their applicability in the classification of the disease with the help of integrative computational methods. Available transcriptomic data in the general population were subjected to the analysis in order to identify important DEGs at the level of statistical providers such as false discovery rate (FDR) adjusted values. The functional enrichment analysis, which consists of Gene ontology (GO) and Kyoto Encyclopaedia of Genes and Genomes (KEGG) pathway analysis was conducted to determine the biological processes and signalling pathways involved in neurodegeneration. Predictive modelling methods, such as L1/L2-regularised Logistic Regression, Random Forest, XGBoost, Support Vector Machine radial basis function kernel and a hybrid quantum-deep learning model, were then used to analyse the identified gene signatures. Notable DEGs were highly enriched with pathways, which dealt with neuroinflammation, synaptic transmission, mitochondrial dysfunction, and dopaminergic signalling. Analysis of comparative classification revealed that models had strong predictive performances, with the integrative hybrid structure having a better discriminative capacity in the form of better accuracy and area under ROC curve (AUC) than baseline strategies. These results should put into the limelight important transcriptomic phenotypes underlying PD and AD and show how integrative modelling schemes could further improve molecular-based disease prediction. The readings of the identified biomarkers and enriched pathways can help in better early diagnosis and give ideas on the targeted therapeutic strategy in neurodegenerative disorders.