324P - Application of advanced machine learning techniques to improve prognosis in primary breast angiosarcoma

Background

Primary Breast angiosarcoma (PBA) is a rare cancer with a gradually increasing incidence and challenging prognosis. This study aimed to develop a machine learning (ML)-based prognostic model that can improve the accuracy of survival rate prediction for patients with PBA.

Methods

The SEER database provided the data used for this study’s analysis (2000–2020). Patients who met any of the following criteria were excluded: diagnosis not confirmed by histology, previous history of cancer or other concurrent malignancies, or unknown data. To identify prognostic variables, we conducted Cox regression analysis and constructed prognostic models using five ML algorithms to predict the 5-year survival. Patient records were randomly divided into training (70 %) and validation (30 %) sets. A validation method incorporating the area under the curve (AUC) of the receiver operating characteristic curve was used to validate the accuracy and reliability of the ML models.

Results

The study population comprised 311 patients. Among them are 135 patients with “surgery,” 51 patients with “surgery + adjuvant radiotherapy (RT)”, 65 “surgery + chemotherapy”, and 60 “surgery + adjuvant RT + chemotherapy”. The therapeutic groups showed significant differences in survival rates with adjuvant RT in combination with surgery, with the highest survival rate (77.1% overall 5 year survival). Multivariate Cox regression analysis revealed laterality and marital status to be significant prognostic factors. ML models revealed that the gradient boosting classifier (GBC) accurately predicted the outcomes, followed by the random forest classifier (RFC), multilayer perceptron (MLP), K-nearest neighbor (KNN), and logistic regression (LR) models. The most significant prognostic factor was the primary site, followed by year of diagnosis and laterality.

Conclusions

These findings underscore the efficacy of adjuvant RT. The integration of ML techniques in this study provides a better understanding of PBA and contributes significantly to personalized medicine.

Clinical trial identification

Editorial acknowledgement

Legal entity responsible for the study

The authors.

Funding

Has not received any funding.

Disclosure

All authors have declared no conflicts of interest.