The first Catalan-specific voice recognition model for healthcare has been developed

With the aim to enhance precision and efficiency in medical documentation, BHH member Hospital Clínic de Barcelona, in collaboration with VÓCALI, has developed the first voice recognition model specifically designed for healthcare in catalan. This innovation, focused on radiology and anatomical pathology, underscores the importance of having language models adapted to the healthcare sector and the linguistic context of Catalonia.

“This project further demonstrates the Hospital Clínic de Barcelona's commitment to technological innovation, aimed at improving its activities while being sensitive to social aspects, such as the use of catalan. It is designed to facilitate the work of healthcare professionals and improve patient care. The project’s main challenge was obtaining a sufficient volume of audio data in catalan, which we were able to achieve through a voice donation campaign within the hospital. Medical staff, retired doctors, and university students all contributed by providing the necessary recordings,” highlighted Alejandro Nieto, Project Manager at Hospital Clínic de Barcelona.

This breakthrough allows healthcare professionals in Catalonia to work in their native language, enhancing comprehension and reducing transcription errors in medical reports. It not only optimizes healthcare workflows but also directly impacts patient care quality by ensuring that medical documentation accurately reflects clinical observations.

“The development of specialized language models is crucial to ensuring the accuracy and relevance of technology in the healthcare sector. Large, general-purpose language models are not suitable for highly specific fields like healthcare, where the technology needs to be tailored to the use case and specific linguistic context,” commented Pedro Vivancos, Director of Innovation Strategy at VÓCALI, the developer of INVOX Medical, the dictation and transcription solution that integrates these new models.

Hospital Clínic de Barcelona: a leader in technological innovation

Hospital Clínic de Barcelona, renowned for its commitment to innovation and excellence, led this project by providing a set of anonymized audio recordings and reports for the voice recognition model training. This collaboration resulted in the creation of a pioneering model that processes catalan with unprecedented accuracy, tailored to the specific needs of the catalan healthcare environment.

The new voice recognition model has been developed specifically for radiology and anatomical pathology. In these areas, healthcare professionals have long used speech-to-text solutions to speed up report generation, allowing them to devote more time to analysis and diagnosis. In the second phase of the project, new models will be developed to cover other medical specialties.

The importance of public-private collaboration in driving innovation

To achieve excellence in our healthcare system, collaboration between the public and private sectors is essential. The public sector possesses the data sets and understands the inefficiencies that need to be addressed through technology. On the other hand, the private sector has the technological expertise required to transform this data into practical models and systems for healthcare professionals' everyday use.

This synergy aims to ensure high-quality healthcare, directly benefiting patient health outcomes. The implementation of these catalan-language voice recognition models will enable professionals at Hospital Clínic de Barcelona, and other hospitals across Catalonia, to produce more accurate and detailed medical reports. This not only optimizes internal workflows but significantly improves the quality of patient care by reducing the time needed to generate diagnostic reports.