Objective To develop a machine learning-based predictive model for cancer-related fatigue (CRF) in elderly breast cancer patients undergoing chemotherapy. MethodsA retrospective analysis was conducted on 366 elderly breast cancer patients who received chemotherapy at the First Affiliated Hospital of Kunming Medical University from January 2022 to December 2025. Patients were divided into a CRF group and a non-CRF group based on CRF status. LASSO (least absolute shrinkage and selection operator) regression was used for variable selection. Three predictive models were constructed using logistic regression (LR), random forest, and support vector machine. Model performance was evaluated, and the optimal model was presented as a Nomogram. ResultsLASSO regression identified seven variables: TNM stage, nutritional risk screening (NRS) 2002 score, short physical performance battery (SPPB) score, albumin (ALB), interleukin-6 (IL-6), C-reactive protein (CRP), and tumor necrosis factor-α (TNF-α). Among the three models, LR achieved the highest area under the curve value [0.893 (0.861, 0.924)] and F1-score (0.806). TNM stage Ⅲ (OR=4.222) and Ⅳ (OR=4.270), NRS 2002 score (OR=2.972), SPPB score (OR=1.345), levels of IL-6 (OR=1.700), CRP (OR=1.795), TNF-α (OR=1.819) were identified as independent risk factors (P＜0.05), and ALB level was identified as independent protective factor (OR=0.899, P=0.004). The calibration curve of the LR-based Nomogram closely approximated the ideal line, with a concordance index of 0.879 and an overfitting degree of 1.07%. ConclusionsCRF in elderly breast cancer patients undergoing chemotherapy is influenced by TNM stage, NRS 2002 score, SPPB score, levels of ALB, IL-6, CRP, and TNF-α. The Nomogram developed in this study demonstrates favorable clinical utility, though further validation is warranted.