Researchers at Ghana's Kwame Nkrumah University of Science and Technology have developed an Artificial Intelligence-assisted model to support Ghanaian Sign Language recognition and translation in healthcare. The innovation, called 'SignTalk-Gh', provides a curated dataset for healthcare, capturing doctor-patient conversations. This development aims to address communication barriers faced by people with hearing and speech impairments in Ghana. The project was a collaborative effort among various departments at the university.
The 'SignTalk-Gh' dataset contains over 9,000 video samples and accompanying metadata, including sentence IDs, English translations, and thematic categories. This vocabulary covers critical terminology such as anesthesia, cardiovascular, and symptoms. According to Dr. Emmanuel Ahene, the Lead Researcher, the dataset is robust for training AI models in complex healthcare scenarios. The system uses advanced deep learning technologies to translate common healthcare terms and phrases.
The development of 'SignTalk-Gh' involved close collaboration among healthcare providers, AI specialists, and members of the deaf community. This ensured that the dataset was culturally relevant, linguistically accurate, and applicable to real-world healthcare environments. The project's goal is to drive inclusive healthcare, especially in underrepresented communities. The researchers noted that despite AI advancements in natural language processing, sign languages, particularly African ones, remain vastly underrepresented.
The disparity in AI research for sign languages is especially critical in healthcare, where effective communication directly affects the quality and outcomes of care for deaf people. Most existing AI-driven systems for sign language recognition focus on American or European sign languages, leaving African sign languages at the periphery of AI research. The 'SignTalk-Gh' dataset could support the future development of recognition and translation systems.
The project was carried out under the auspices of the Responsible Artificial Intelligence Lab and the Artificial Intelligence for Sustainable Development project. It was funded by the French Embassy in Ghana, the Agence Française de Développement, and other organizations. The researchers said the dataset demonstrated suitability for retrieval-based text-to-sign applications.
The development of 'SignTalk-Gh' highlights the potential of AI in driving inclusive healthcare, especially in underrepresented communities. According to Dr. Ahene, the project has a clear roadmap for model training, deployment, and future expansion. This innovation has the potential to address pressing societal challenges in communication.
The 'SignTalk-Gh' dataset serves as a powerful resource for addressing communication barriers faced by people with hearing and speech impairments in Ghana. The project's success could pave the way for similar initiatives in other African countries, promoting inclusive healthcare through AI technology.
Key points
- The 'SignTalk-Gh' dataset contains over 9,000 video samples and accompanying metadata for Ghanaian Sign Language recognition in healthcare.
- The project was a collaborative effort among various departments at Kwame Nkrumah University of Science and Technology.
- The development of 'SignTalk-Gh' aims to drive inclusive healthcare, especially in underrepresented communities.
