[Perspectives on Device Selection in Japan:Artificial Intelligence Utilization, Distribution Systems, and Future Challenges]

Abstract

Device selection in neuroendovascular therapy has traditionally relied on operator experience and institutional routines, leading to variability and limited reproducibility. Advances in artificial intelligence (AI), particularly machine learning, now allow analysis of complex clinical and morphological data to support optimal device selection. AI can visualize and share tacit expert knowledge, enhance decision-making, and promote procedural standardization and education. However, its recommendations remain probabilistic, and final clinical responsibility must rest with the physician. Reliable application requires adequate multi-institutional data and improved interpretability. In addition to supporting clinical judgment, AI-based prediction can optimize medical device logistics by forecasting necessary items and reducing excessive deliveries and returns. Such efficiency is increasingly important in the context of workforce shortages and transportation constraints, including Japan's "2024 logistics problem." We developed an integrated system combining AI-driven device selection with an electronic ordering platform, enabling real-time information sharing among physicians, distributors, and logistics providers. In preliminary validation involving unruptured aneurysm cases, this approach reduced delivered items by 20-55% without compromising safety. AI-assisted device selection therefore represents both a clinical advancement and a pathway toward more sustainable medical supply chains, linking healthcare, industry, and logistics through data-driven optimization.