Development and evaluation of a predictive model for postoperative recurrence and metastasis in breast cancer using an artificial intelligence ultrasound breast system.

Abstract

Objective: To assess the feasibility and efficacy of developing a predictive model for postoperative recurrence and metastasis in breast cancer using the Artificial Intelligence Ultrasound Breast System (AIUBS).

Methods: A retrospective study was conducted with 120 breast cancer patients who underwent surgery between January 2022 and December 2023. Patients were divided into two groups based on postoperative outcomes: recurrence/metastasis (n = 58) and non-recurrence/non-metastasis (n = 62). Logistic regression was used to identify independent predictors, and a nomogram model was constructed. Model performance was assessed using Receiver Operating Characteristic curves, calibration curves, and decision curve analysis (DCA). The optimal cutoff value was determined through confusion matrix analysis.

Results: Univariate analysis identified lymph node metastasis (OR = 8.17, 95% CI: 3.51-18.99), estrogen receptor (ER) status (OR = 0.46, 95% CI: 0.21-0.99), and human epidermal growth factor receptor 2 status (OR = 5.32, 95% CI: 2.32-12.22) as significant predictors. Multivariate analysis confirmed lymph node metastasis (OR = 8.81, 95% CI: 3.68-21.07) and ER status (OR = 0.39, 95% CI: 0.16-0.94) as independent predictors. The nomogram model demonstrated an Area Under the Curve of 0.77 (95% CI: 0.68-0.85). The optimal cutoff value, derived from confusion matrix analysis, was 0.572, confirming the model's clinical utility.

Conclusion: The AIUBS-based predictive model for postoperative recurrence and metastasis in breast cancer demonstrates high predictive accuracy and clinical utility, providing valuable support for personalized treatment and follow-up decisions.