Harmonising linguistic validation with AI: Precision, efficiency, and the human touch in patient-reported outcome translation

Abstract

The surge in artificial intelligence (AI) prompts a reassessment of linguistic validation methods for patient-reported outcome (PRO) measures. The robust linguistic process, designed to adapt PRO measures for different cultures and languages, is upheld by regulators and the outcomes research community for its value in maintaining concept equivalence across global trial data. Its methods are entrenched in human translation and review, making it more challenging to integrate AI (machine learning, deep learning, natural language processing) compared to other parts of the global localisation industry. This article provides an overview of the key challenges in integrating linguistic validation and AI. Despite these hurdles, it advocates for the industry to embrace the potential benefits through collaborative and responsible innovation.