On the Design of a Sign Language Corpus of Medical Terms for Automatic Translation Systems: Mixed Methods Approach

基于混合方法的医学术语手语语料库自动翻译系统设计

阅读:1

Abstract

BACKGROUND: Hearing loss is a global health issue affecting millions and creating significant communication barriers, particularly in accessing health care services. These barriers can lead to complications and iatrogenic events, emphasizing the need for assistive technologies that enhance communication efficiency. OBJECTIVE: This study aimed to develop a corpus of medical terms for the "Captar-Libras" project, designed to improve communication between health care professionals and deaf patients through a bidirectional sign language system. METHODS: This study used the Delphi method to obtain consensus on key terms for a sign language translation system in health care emergency consultations. Initially, a questionnaire with common emergency questions was developed and distributed to health care professionals. The collected data were analyzed by a team of experts and adapted to Brazilian Sign Language (Língua Brasileira de Sinais [Libras]). Simulated clinical scenarios were then created to validate the system and ensure the accuracy of the vocabulary in the medical context. RESULTS: Among the 16 participants, most were physicians (n=14, 87.5%) with experience in emergency care, and half had previously treated patients with hearing loss in emergency settings. The questions evaluated received high average importance scores, particularly those related to initial symptoms and pain intensity. Some suggestions for adjustments were made, with two wording modifications significantly improving clarity regarding smoking and alcohol use. Additional suggestions to enhance the medical interview were also proposed. This study aimed to identify essential questions for emergency consultations with deaf patients, focusing on developing a corpus for Libras recognition system. The findings emphasize the importance of effective communication and highlight the challenges of translating medical terms into Libras. To address these complexities, a multidisciplinary team used the Delphi method to ensure linguistic and cultural accuracy. Additionally, the study reinforces the need for clear, structured medical queries to improve accessibility in emergency care. As a next step, system validation through simulated scenarios will be conducted. Despite certain limitations, this research lays a solid foundation for advancing sign language recognition in medical settings. CONCLUSIONS: This study represents a significant methodological step toward improving communication between health care professionals and deaf individuals in emergency medical settings. Rather than proposing a universal solution, the study presents a structured and participatory approach for developing a corpus of medical terms in Libras, with interdisciplinary validation. The process included the involvement of deaf sign language experts during the translation and linguistic adaptation phase, ensuring that the corpus reflects authentic usage and articulation in Libras.

特别声明

1、本页面内容包含部分的内容是基于公开信息的合理引用;引用内容仅为补充信息,不代表本站立场。

2、若认为本页面引用内容涉及侵权,请及时与本站联系,我们将第一时间处理。

3、其他媒体/个人如需使用本页面原创内容,需注明“来源:[生知库]”并获得授权;使用引用内容的,需自行联系原作者获得许可。

4、投稿及合作请联系:info@biocloudy.com。