IN a groundbreaking study, artificial intelligence (AI) and machine learning (ML) have been utilised to predict local recurrences (LR) and distant metastasis (DM) in breast cancer patients. The research, which analysed data from 154 patients over an average follow-up of 133.16 months, marks a significant step towards improving the prognostic capabilities of breast cancer care.
Traditionally, predicting recurrence location has been challenging. LRs can manifest in the breast parenchyma, chest wall, skin, or surgical scar tissue. However, the study revealed that ML models could effectively identify LR localisation. With a model accuracy (ROC AUC) of 0.75 for predicting LR in breast tissue and 0.69 for surgical scar tissue, the results are promising.
Additionally, the study found that the same models could predict DM risk after recurrence, achieving an accuracy of 0.74. Key factors influencing these predictions included recurrence localisation, time elapsed since the initial diagnosis, and adjuvant chemotherapy treatments.
The study’s findings emphasise the potential of combining traditional clinical factors with AI-driven insights to offer a more precise risk assessment for patients. The integration of machine learning into oncological decision-making could transform how clinicians approach treatment planning, monitoring, and follow-up care, ultimately leading to better outcomes for breast cancer patients.
Helena Bradbury, EMJ
Reference
Kovács KA et al. Machine learning prediction of breast cancer local recurrence localization, and distant metastasis after local recurrences. Scientific Reports. 2025;15:4868.
