Abstract:Background and Aims Breast-conserving surgery, which balances tumor excision with preservation of breast tissue, has become a widely adopted surgical approach for breast cancer. However, postoperative tumor recurrence remains a major factor affecting patient prognosis. Accurate risk prediction tools are urgently needed to guide personalized treatment strategies. This study aimed to develop a risk prediction model for tumor recurrence after BCS based on admission clinical data and to evaluate its predictive performance to provide a scientific and practical tool for clinical decision-making.Methods A total of 224 breast cancer patients who underwent breast-conserving surgery between May 2017 and May 2019 were enrolled. Postoperative recurrence was recorded during follow-up. Multivariate Logistic regression analysis was used to identify independent risk factors for recurrence and to construct a risk prediction model. The model's discriminative ability was assessed using the receiver operating characteristic (ROC) curve, and its calibration was evaluated using a calibration curve.Results 208 patients completed follow-up, ranging from 32 to 84 months, with a mean duration of (58.41±7.33) months. The recurrence rate was 17.79%. Multivariate Logistic regression analysis revealed that TNM stage Ⅲ (OR=2.029), tumor diameter ≥4 cm (OR=1.782), ≥4 lymph node metastases (OR=1.958), lymphovascular invasion (OR=1.984), and HER2 positivity (OR=1.774) were independent risk factors for recurrence (all P<0.05). The Logistic regression model was established as follows: Y=-12.788+0.707X?+0.578X?+0.672X?+0.685X?+0.573X?. The model yielded an area under the ROC curve (AUC) of 0.934 (95% CI=0.891-0.963), with a sensitivity of 86.49% and specificity of 96.49%. The calibration curve demonstrated good agreement between predicted and observed outcomes (χ2=0.501, P=0.392).Conclusion TNM stage Ⅲ, tumor diameter ≥4 cm, ≥4 lymph node metastases, lymphovascular invasion, and HER2 positivity are independent risk factors for tumor recurrence after breast-conserving surgery. The risk prediction model based on these factors demonstrates favorable discrimination and calibration, offering valuable guidance for postoperative risk assessment and clinical intervention.