ESTRO 2025 - Abstract Book

S3767

Physics - Radiomics, functional and biological imaging and outcome prediction

ESTRO 2025

Figure 2 : Decision tree for the Gini-based model method. IL8: interleukin-8; LYMPH: lymphocyte count; HEM : hemoglobin

Conclusion: The study demonstrated the potential of decision tree models to provide explainable multiclass predictions of mortality risk in patients undergoing PRT for bone metastases. IL8, lymphocyte count, and hemoglobin emerged as critical predictors of survival, underscoring the prognostic value of laboratory data in palliative care. This approach supports personalized treatment planning and enhances clinical decision-making by offering transparent and understandable predictions of life expectancy.

Keywords: Machine learning , radiotherapy, advanced cancer