Persian language comprehensibility in machine translation vs. human translation in health-related multilingual context based on the Patient-oriented and Culturally-adapted (POCA) model

Abstract

Accessibility to healthcare information is contingent upon effective health communication, particularly for individuals with poor English ability. Communication is hampered by language problems, which can lead to miscommunication and restricted access to healthcare. While machine translation (MT) presents a potential solution, its effectiveness in healthcare remains understudied. This study assesses the linguistic comprehensibility of health-related translations using both MT and human translation (HT), aiming to identify challenges and opportunities for enhancing health communication. The study utilised mixed methods, combining qualitative thematic analysis with quantitative poll questionnaires. Health-related translations from the ‘Health Translations’ website were categorised and assessed for linguistic comprehensibility using comprehensibility using the Patient-oriented and Culturally-adapted (POCA) model's six categories. MT generally outperformed HT in generating high-quality translations, especially in technical or specialised terminology and abstract concepts. However, HT excelled in categories requiring nuanced language and context comprehension. Specific linguistic challenges identified included abstract translations, coherence issues, and complex sentence structures, hindering comprehension among non-native speakers. Despite challenges, MT showed significant progress in improving linguistic comprehensibility, particularly in health-related contexts. This study provides insights into MT and HT effectiveness in health communication. In healthcare settings, using precise language is essential to improving comprehension and minimising misinterpretation. © 2025 Informa UK Limited, trading as Taylor & Francis Group.