Advancing BNCT Precision Medicine through Accelerated Aptamer Selection and AI-Driven Data Mining for High-Sensitivity Boronophenylalanine Monitoring

Abstract

The advancement of Boron Neutron Capture Therapy (BNCT) requires rapid and accurate monitoring of Boronophenylalanine (BPA), yet existing analytical methods such as ICP-AES and ICP-MS remain too slow and incompatible with real-time clinical needs. To address this limitation, we developed an integrated aptamer-discovery and sensing framework that combines capture SELEX, Next-Generation Sequencing (NGS), and artificial intelligence–enhanced data mining, followed by validation on a field-effect transistor (FET) biosensing platform. At the molecular selection level, a redesigned capture SELEX protocol was developed to effectively accommodate compact small molecules such as BPA without requiring immobilization on solid substrates. This design minimizes selection bias and preserves native binding configurations. High-resolution NGS analysis, together with AI-assisted motif clustering and structure–function modeling, enabled the identification of aptamer families exhibiting strong and selective affinity toward key chemical domains of BPA, yielding insights beyond the capabilities of classical enrichment-based approaches. Aptamers identified through this SELEX-NGS-AI pipeline were subsequently integrated onto an extended-gate FET biosensor, which demonstrated clear, reproducible electrical responses with a well-defined detection limit. This confirms that the combined workflow can efficiently generate high-performance small-molecule aptamers and translate them directly into electronic biosensing platforms suitable for real-time BPA detection in BNCT. Overall, this work establishes a unified and synergistic pipeline in which capture SELEX provides the basis for molecular recognition, NGS offers high-resolution sequence evolution tracking, and AI converts complex sequence patterns into predictive functional insights. The resulting BPA-specific aptamers enabled sensitive, label-free electrical detection on an FET sensor, demonstrating a scalable and clinically relevant route toward high-precision BPA quantification for BNCT monitoring. Figure 1