Supportive and Palliative Care Indicators Tool

The Korean team began their translation by using Large Language Models (LLMs), such as Open AI’s ChatGPT and Google’s Gemini, to generate the initial forward and back translations of SPICT. These AI-generated translations were subsequently evaluated using both standardised automated linguistic quality metrics and human expert assessments, ensuring a robust foundation for clinical validation and cultural adaptation.

Following the initial translation phase, a panel of eight experts in palliative care, nursing, instrument development, and natural language processing (NLP) conducted a rigorous three-round Delphi consensus process. This included open-ended feedback, iterative revision of each item, and evaluation of clinical and cultural validity. A final in-person consensus meeting was held to refine language tone, grammar, and terminological consistency.

The application of advanced AI technology helped expedite the translation process and offered a useful baseline for human expert review. Importantly, we have clearly documented the use of LLMs and ensured that the final version of SPICT-KR was thoroughly validated by clinicians.

The study on the development and validation of the Korean version of the SPICT 2025 using LLMs is currently being prepared for publication in a peer-reviewed academic journal. In addition to disseminating the methodological findings, the research team plans to conduct applied studies using the finalised SPICT-KR. These future studies will focus on the practical implementation of the SPICT-KR to assess palliative care needs and facilitate the early identification of target populations among healthcare professionals and patients in the Republic of Korea. The results are expected to contribute to enhancing palliative care delivery and improving patient outcomes in diverse clinical settings.