A Scoping Review of Malnutrition Risk Prediction Models in Cancer Patients

OBJECTIVE: To systematically review the existing literature on risk prediction models for malnutrition in cancer patients. BACKGROUND: Malnutrition is highly prevalent among cancer patients, and timely identification and intervention can improve patient outcomes and quality of life. Risk prediction models can forecast the future disease risk of patients. This study aims to conduct a comprehensive scoping review of malnutrition risk prediction models for cancer patients developed domestically and internationally, analyse current gaps in the field, and provide references for clinical practice and future research. METHODS: This review followed the PRISMA Extension for Scoping Reviews and the Arksey and O'Malley framework, and the study was conducted in accordance with the CHARMS checklist. Systematic searches were conducted in PubMed, Web of Science Core Collection, the Cochrane Library, EMbase, CINAHL, CNKI, Wanfang, and Sinomed, with the search period spanning from database inception to October 23, 2025. Original studies reporting the development or validation of malnutrition risk prediction models in cancer patients were eligible for inclusion. RESULTS: A total of 36 studies were included in this scoping review. Extracted data included study characteristics, model development methods, predictors included, and performance evaluation metrics. Considerable overlap was observed in the development methods across models, with traditional statistical analysis being predominant. Retrospective data collection was applied in the majority of the included studies. These prediction models covered multiple malignancies, including rectal, liver, nasopharyngeal, gastric, and oesophageal cancers. A wide range of predictors was used, with age and BMI being the most frequently included. CONCLUSION: This review maps cancer patients' malnutrition prediction models. Existing models screen malnutrition but have flaws in construction and validation. Future studies will adopt longitudinal, multimodal data and AI to optimize tools for early screening and intervention.