<p>This study presents an integrated approach combining photoelectrochemical (PEC) biosensing, density functional theory (DFT), and machine learning (ML) to address diagnostic and mechanistic challenges in cholangiocarcinoma (CCA). A novel Bi₄Ti₃O₁₂/WS₂ heterojunction was engineered as a high-performance PEC platform, exhibiting broad-spectrum absorption extending to 850&#xa0;nm and a 70% reduction in photoluminescence intensity due to suppressed electron–hole recombination. The optimized biosensor achieved ultrasensitive miR-29a detection with a linear range of 1 fM–200&#xa0;nM, a detection limit of 0.19 fM, and 98.2–102.5% recovery in spiked serum, alongside robust specificity and reproducibility (1.3% RSD over 16 cycles). DFT calculations revealed interfacial electronic coupling between Bi₄Ti₃O₁₂ and WS₂, narrowing the bandgap to 1.571&#xa0;eV and establishing a type-II charge transfer pathway, which synergistically enhanced photocurrent generation compared with pristine WS₂. Parallel ML analysis of multi-omics datasets uncovered consistent miR-29a upregulation in CCA tumors and identified integrins COL4A1/CDK6 as top discriminative targets. The Random Forest classifier integrating miR-29a and targets achieved 0.890 AUC for tumor diagnosis, revealing pathway-specific regulatory networks. The bridging nanomaterial innovation with computational biology not only advances ultrasensitive miRNA detection but also deciphers miR-29a’s dual role in CCA pathogenesis, offering a novel framework for precision oncology.</p> Graphical Abstract <p></p>