Addressing Clinical Documentation Challenges: The Role of AIPowered Digital Scribes

Abstract

In the realm of healthcare, the burden of clinical documentation has long been a challenge for healthcare professionals. This paper presents a comprehensive study on the development and evaluation of a unified digital scribe system aimed at automating the process of recording, transcribing, and annotating clinical narratives in Spanish. The proposed system integrates cutting-edge technologies, including speech recognition and Named Entity Recognition (NER), to streamline the documentation workflow. The study evaluates the performance of the system in a simulated environment, encompassing transcription accuracy and entity recognition across medical and dental domains. Notably, the average Word Error Rates (WER) for different domains reveal promising results, showcasing the system's potential to accurately transcribe spoken clinical narratives. Furthermore, the NER model demonstrates its capability to identify key medical entities such as diseases, body parts, and medications, with F1 scores indicating strong performance. The paper also outlines the development of a prototype platform that unifies both the transcription and NER modules. This platform offers healthcare professionals a seamless solution for capturing, transcribing, and annotating clinical information, ultimately aiming to reduce documentation burden and improve clinical workflow efficiency. In conclusion, this research presents a robust foundation for the development of a digital scribe system in the Spanish healthcare system, offering promising solutions to alleviate clinical documentation challenges, reduce professional burnout, and enhance patient care through streamlined documentation practices.