Background: Cancer remains a leading global health burden. Artificial intelligence offers new opportunities to address complex physical and psychological symptoms in palliative and supportive cancer care. Despite rapid advances, including large language models, these technologies have not been consistently reviewed in this context, highlighting a gap in the synthesised literature. Aim: To map current evidence on how artificial intelligence is implemented in palliative and supportive care for people with cancer and their caregivers, and to identify associated challenges and future directions. Design: Scoping review guided by the Joanna Briggs Institute (JBI) framework. Data sources: We searched PubMed, CENTRAL, PsycINFO, and CINAHL databases, along with grey literature, up to 17 September 2024, and manually screened reference lists. Results: Of 2618 records screened, 199 studies were included. Most applications supported healthcare professionals, mainly for predictive purposes such as prognosis, symptom monitoring, and treatment-related adverse events. Machine learning was the predominant approach, reflecting reliance on structured clinical data. Publications increased markedly after 2020, with a sharp rise from 2022 onwards. Only 26 studies involved direct use by people with cancer or their caregivers, and large language model–based applications remained rare, indicating limited patient-facing use. Conclusions: Artificial intelligence is increasingly being applied in palliative and supportive care, yet applications designed for direct patient or caregiver use remain scarce. Further efforts should prioritise the development and validation of ethically sound, clinically integrated artificial intelligence tools to support person-centred palliative and supportive care.
Kikuchi et al. (Fri,) studied this question.