Cancer-related fatigue (CRF) is the most common and distressing symptom in cancer survivors, severely affecting their quality of life. However, clinicians and patients are not well recognized for its importance and lack timely screening and assessment. With the rapid development of artificial intelligence and personalized care, early screening and assessment of CRF using machine learning to construct risk prediction models may contribute to this. Therefore, we redefined the predictors of CRF based on the predisposing, precipitating, and perpetuating (3P) model to develop a valid basis for the feature selection of future prediction models, intending to provide a more accurate and personalized plan for the clinical diagnosis and management of CRF.