A Personalized Point-of-Care Platform for Discovery and Validation of miRNA Targets Using AI and Edge Computing...

Area of Science:

• Oncology
• Bioinformatics
• Medical Diagnostics

Background:

• Personalized precision medicine is transforming cancer therapy.
• Current diagnostics rely on centralized labs, causing treatment delays.
• Point-of-care (POC) diagnostics are needed to bridge this gap.

Purpose of the Study:

• To introduce GTMIT, a novel AI-powered POC platform for real-time microRNA (miRNA) target discovery and validation.
• To enable immediate therapeutic decision-making at the patient's bedside, bypassing cloud reliance.
• To accelerate cancer treatment initiation by integrating molecular diagnostics with POC interventions.

Main Methods:

• Developed a Transformer-GNN hybrid architecture with Power Normalization for miRNA-mRNA interaction prediction.
• Implemented SNP-adaptive Gapped Pattern Graph Convolutional Networks (GP-GCN) to address patient-specific genetic variations.
• Integrated edge therapeutic optimization considering regional cancer prevalence and resource limitations.

Main Results:

• GTMIT achieved high performance on clinical datasets: 94% AUC, 87% precision, and 79% recall.
• The platform demonstrated effective real-time miRNA target discovery and validation at the POC.
• Edge computing enabled local decision-making without cloud dependency.

Conclusions:

• GTMIT significantly reduces time-to-treatment from days to minutes.
• The platform enhances personalized cancer therapy, especially in resource-limited settings.
• GTMIT bridges molecular diagnostics and immediate intervention for improved patient outcomes.