MACHINE learning models accurately predicted survival outcomes in patients with breast cancer bone metastasis, while also identifying individuals at increased risk of early death, according to a large retrospective analysis.
Researchers analysed data from 23,723 patients diagnosed with de novo Stage IV breast cancer and bone metastasis between 2010 and 2020, and compared multiple machine learning approaches for survival prediction.
Models were subsequently assessed in an independent multicentre cohort of 230 patients.
Among all evaluated algorithms, the Light Gradient Boosting Machine (LightGBM) model demonstrated the most consistent prognostic performance, with decision curve and calibration analyses supporting the model’s potential clinical utility.
The study also focused on early death, defined as death within 3 months of diagnosis.
Approximately 16% of patients in the population-based cohort experienced early death within this timeframe.
Although a stacking model was best at distinguishing between patients who did and did not experience early death, LightGBM delivered the most balanced overall performance across accuracy, precision, specificity, calibration, and clinical utility measures.
A sensitivity analysis excluding surgery, radiotherapy, and chemotherapy variables still showed that LightGBM produced acceptable predictive performance.
According to the investigators, this suggests the model may support risk assessment using information available at diagnosis.
Researchers also explored the association between primary tumour surgery and survival outcomes.
Before matching, surgery was associated with longer survival compared to no surgery, and after propensity score matching, the survival advantage remained significant.
Similar findings were observed in the external cohort.
However, the researchers emphasised that treatment response, performance status, frailty, and metastatic burden were unavailable, meaning residual confounding and selection bias could not be excluded.
As a result, the observed association should be interpreted cautiously and not as evidence of causality.
Future prospective studies incorporating broader clinical data will be needed to validate these models and clarify how they could support decision-making in routine oncology care.
Reference
Fang J et al. Machine learning-based prognosis and early death prediction in de novo stage IV breast cancer patients with bone metastasis: a SEER database and multicentre retrospective study. Sci Rep. 2026;DOI:10.1038/s41598-026-62142-w
Featured image: romaset on Adobe Stock
Each article is made available under the terms of the Creative Commons Attribution-Non Commercial 4.0 License.
