With the aging population and increasing prevalence of chronic diseases, a team led by Yi-Lwun Ho, Director of the National Taiwan University Hospital (NTUH) Telecare Center, has collaborated with Chunghwa Telecom to develop the Intelligent Voice Emergency Prediction System. This system employs multilingual voice recognition to convert medical conversations into high-accuracy text. Using natural language processing (NLP) and semantic analysis, it predicts patients’ emergency risk and adjusts care intensity and prioritization accordingly. This innovation not only enhances case managers' efficiency but also identifies high-risk patients at an early stage, optimizing the overall quality of care.
Dual Challenges: Multilingual Complexity and Lack of Risk Alerts
Current remote care systems can monitor patients' physiological data in real time but face significant challenges in speech recognition. First, medical conversations are complex, often mixing multiple languages and medical terminologies. Conventional speech recognition software struggles to accurately transcribe such content, forcing case managers to spend excessive time on documentation. Additionally, existing systems lack predictive decision-support tools that analyze patients' historical conversations and physiological data to assess emergency risks. As a result, healthcare providers are unable to proactively identify high-risk patients.
Speech Recognition Combined with Risk Prediction: Doubling Care Efficiency
The Intelligent Voice Emergency Prediction System integrates multilingual speech recognition technology, enabling accurate transcription of Mandarin, English, and Taiwanese Hokkien medical dialogues with an 87.22% accuracy rate. This significantly reduces the documentation workload for case managers. Furthermore, NLP technology is applied to analyze transcribed text, which is then cross-referenced with patient nursing records and physiological datato predict emergency risks with 89% accuracy. Through this predictive system, case managers can identify high-risk patients in advance, adjust care intensity and prioritization, and maximize the critical treatment window—ultimately reducing both emergency visits and hospitalization days.
Breaking Language Barriers and Expanding Global Healthcare Access
Yi-Lwun Ho emphasized that the implementation of the Intelligent Voice Emergency Prediction System will greatly enhance the efficiency of remote healthcare. He highlighted its potential to revolutionize Taiwan’s healthcare modeland expand global applications, particularly among Taiwanese expatriates and overseas Chinese communities. This system effectively addresses language barriers and remote healthcare resource shortages. Looking ahead, the system will be extended to more healthcare settings, increasing market adoption, and will be further applied to various disease areas, continuously improving the quality of healthcare worldwide.
Resource: 智慧辨識語音結合生理數據 遠距預測急診風險調整照護
