Gene-Based Predictive Modelling for Enhanced Detection of Systemic Lupus Erythematosus Using CNN-Based DL Algorithm

Area of Science:

• Immunology
• Computational Biology
• Genomics

Background:

• Systemic Lupus Erythematosus (SLE) is a complex autoimmune disease with variable clinical presentations.
• Current diagnostic methods for SLE often lack sufficient sensitivity and specificity, leading to diagnostic delays.
• Advanced computational approaches are needed to improve the accuracy and timeliness of SLE diagnosis.

Purpose of the Study:

• To develop and evaluate a novel gene-based predictive model for diagnosing Systemic Lupus Erythematosus (SLE).
• To leverage Stacked Deep Learning Classifiers (SDLC) trained on transcriptomic and clinical data for enhanced SLE identification.
• To assess the performance of the SDLC model against traditional diagnostic approaches.

Main Methods:

• Utilized transcriptomic data from the Gene Expression Omnibus (GEO) database.
• Integrated gene expression data with clinical features and laboratory results.
• Developed and trained Stacked Deep Learning Classifiers (SDLC), including SBi-LSTM and ACNN models.

Main Results:

• The SDLC model achieved a diagnostic accuracy of 0.996 for SLE.
• Individual deep learning models, SBi-LSTM and ACNN, demonstrated accuracies of 92% and 95%, respectively.
• The ensemble learning approach of SDLC effectively identified complex patterns in multi-modal data.

Conclusions:

• Deep learning methods, particularly SDLC, show significant potential for improving SLE diagnosis.
• Integration of open-access data repositories like GEO with advanced computational models can advance SLE management.
• This research highlights the promise of precision medicine in the diagnosis and treatment of SLE.