Application of Artificial Intelligence (AI) in cancer symptom management for adult cancer survivors: a scoping review

Tao Wang; Haiying Wang; Mengyuan Li; Daniel Bressington; Daniel Terry; Guanjin Wang; Wai Hang Kwok; Jing-Yu (Benjamin) Tan

doi:10.1016/j.ijmedinf.2026.106643

Highlights • Artificial Intelligence (AI) supports cancer symptom management through detection, monitoring, decision support, and patient-facing education/counselling. • Common AI approaches include natural language processing (NLP), machine learning, and conversational AI, using Electronic Health Records data for AI model development and patient-reported data for AI-enabled intervention implementation. • Moderate-to-high performance was reported in most AI model development studies and improvements in clinical and psychosocial outcomes in most AI-enabled interventional studies. • Identified key influencing factors include technical performance, data quality, workflow integration, usability/equity, communication, and ethics. • Existing evidence remains early-stage and context-dependent; future research should emphasise validation, co-design, and real-world, equitable implementation. Abstract Background and aim Artificial Intelligence (AI) has been increasingly used in cancer survivorship to support symptom management. This scoping review aimed to map existing evidence on AI applications in cancer symptom management for adult cancer survivors, including AI model development, AI-enabled intervention delivery and adoption, symptom targets, key features of the AI approaches used, reported outcomes, influencing factors, and research gaps to inform future priorities. Method This scoping review was conducted in accordance with the Joanna Briggs Institute methodology for scoping reviews. Eight electronic databases were comprehensively searched in November 2025 and updated in March 2026. Empirical studies published in English from 2015 onward that focused on the development and/or implementation of AI models for cancer-related symptom care among adult survivors were included. Given the heterogeneity of the included studies, findings were synthesised using descriptive statistics and narrative analysis. Factors influencing AI development and implementation were analysed using inductive content analysis. Results A total of 41 studies were included: 21 focused on AI model development, 18 on AI-enabled intervention delivery, and 2 on both. Common techniques included natural language processing, machine learning, and conversational AI. Development studies primarily used unstructured electronic health record data, whereas delivery studies more often relied on patient-reported inputs. AI applications were mainly used to support symptom detection (n = 15), monitoring (n = 5) and triage (n = 2); clinical decision support (n = 3) and personalised management (n = 5); and patient education (n = 9) or counselling (n = 2). Nearly half of the studies (n = 18) addressed general or multiple-symptom management, while others focused on specific symptoms, particularly pain (n = 9) and psychological distress (n = 9). Model performance was generally moderate to high, varying by symptom, task complexity, and data source. Delivery studies reported improvements in clinical and psychosocial outcomes. Six categories of influencing factors were identified: technical performance; data quality, documentation, and generalisability; clinical workflow integration; usability, access, and equity; human-centred communication; and ethics and safety. Conclusion AI is increasingly used to support symptom management in adult cancer survivors, primarily through symptom identification, monitoring, decision support, and patient-facing recommendations, education and counselling. However, the evidence remains heterogeneous, early-stage, and context-dependent, with variability in AI functionality, technical reporting, study populations, healthcare settings and study designs. Future research could prioritise clearer reporting of AI functionality, broader symptom coverage, robust validation, co-design, and clinically integrated evaluations that demonstrate safety, equity, effectiveness, and real-world usefulness beyond predictive accuracy.

Application of Artificial Intelligence (AI) in cancer symptom management for adult cancer survivors: a scoping review

Files and links (1)

Abstract

Details

UN Sustainable Development Goals (SDGs)

Metrics