Multilingual dialogue systems using neural networks for apps in the healthcare domain: the triage (Spanish-English/Arabic)

Ref. no. UMA18-FEDERJA-067
2019-2022


The Triage project falls within interpreting for the public services. The interpreting industry is now experiencing a rapid digitalisation and technologicalisation process. Specifically in the field of public services, so-called remote interpreting is beginning to be introduced, which allows the service to be offered by telephone or video conference software. This modality reduces the cost of interpreting, although it also requires a high investment on human interpreters. In the field we are concerned with (healthcare), effective communication with the patient is essential for adequate and quality care. However, since interpreting services are expensive, not all hospitals and health centres can afford to treat foreign patients in their own language. This situation is particularly complex in the case of the Andalusian health system, due to migratory movements and the volume of tourists visiting us.

The ultimate goal of the project is the development of a multilingual system to automate triage. The term ‘triage’ refers to the process by which people are selected based on their need for immediate medical care when available resources are limited. We will focus on emergency triage, as it is the scenario that requires the fastest reaction time and has the least time to resort to external interpretation services. Our main objective is the design and implementation of a multilingual system to enable effective communication between the healthcare professional and the patient which will allow patients to be assessed and ordered according to the severity and urgency of the case. The system is based on multilingual dialogical models and multimodal neural automatic translation (speech-text-speech). The system allows automatic translation/interpretation between Spanish and English or Arabic. The methodology used combines various Natural Language Processing techniques, such as neural-based machine translation, machine learning, speech recognition (speech-text) and synthesis (text-to-speech), in addition to translation and interpretation, corpus linguistics, digital resources and multilingual linguistic technologies.