In the high-pressure environment of emergency departments, physicians often juggle patient consultations while simultaneously entering medical records, and nurses are frequently required to work overtime to complete documentation after their shifts. This demanding daily routine highlights a critical area where smart healthcare solutions are urgently needed.
To help relieve healthcare workers’ stress and improve care efficiency, the Industrial Technology Research Institute (ITRI) has partnered with Far Eastern Memorial Hospital, Chung Shan Medical University Hospital, Kaohsiung Veterans General Hospital, and Chi Mei Medical Center to launch the MedBobi 2.0 integrated platform. Centered on AI, the platform has evolved from simple speech recognition into a clinical decision support assistant capable of understanding medical context and providing real-time responses, marking a new phase of intelligent collaboration. The goal is to build a national team for medical AI and inject strong momentum into a healthier Taiwan.
In response to Taiwan’s challenges of nursing workforce shortages and nurse-to-patient ratios far exceeding international recommendations, ITRI developed MedBobi 1.0 “VoiceEasy for Healthcare” in 2024. After its implementation at Far Eastern Memorial Hospital, it used speech recognition to transcribe patient interviews in real time, reducing nurses’ documentation workload by about 90%.
However, the development team found that for AI medical applications to be successfully implemented, abundant training “corpora” (commonly used clinical terms) and a compatible, scalable system are essential. In Taiwan, the diversity of professional terminology across different medical systems and specialties has made it difficult to standardize nationwide. In addition, variations in hospital software specifications and IT infrastructures have hindered cross-institution integration and international alignment.
To address these challenges, the team adopted an open integration strategy and upgraded the system into MedBobi 2.0. It connects clinical corpora from four major medical systems across northern, central, and southern Taiwan, covering local dialects such as Mandarin, Taiwanese, and Hakka. It also incorporates the international standard FHIR (Fast Healthcare Interoperability Resources) to create an extensible and interoperable smart healthcare platform.
Leveraging Agentic AI technology, the system not only understands medical terminology and clinical contexts but can also proactively provide diagnostic and medication suggestions, transforming from a simple recording tool into an intelligent assistant with clinical decision-making capability.
By integrating multi-institution and multi-specialty clinical corpora, MedBobi 2.0 enables real-time consultation documentation, handover summary generation, and nursing record suggestions, significantly reducing administrative workload and improving overall care quality. Through its open platform architecture and collaborative industry partnerships, it is paving the way for Taiwan to establish standardized, scalable models for implementing AI in healthcare. Looking ahead, the system can be extended to long-term care facilities and even exported to New Southbound countries such as Thailand and Vietnam, positioning itself as a flagship example of Taiwan’s smart healthcare innovation and a powerful tool for international outreach.
Resource: 創新平台／智慧協作 打造醫療AI國家隊
