1. Treatment-related features improve machine learning prediction of prognosis in soft tissue sarcoma patients
- Author
-
Michael Bernhofer, Jan C. Peeken, Fridtjof Nüsslin, Francesco Pasa, Christoph Knie, Kerstin A. Kessel, Burkhard Rost, Stephanie E. Combs, Basil Komboz, Tatyana Goldberg, Pouya D. Tafti, and Andreas E. Braun
- Subjects
Adult ,Male ,medicine.medical_treatment ,Machine learning ,computer.software_genre ,Risk Assessment ,030218 nuclear medicine & medical imaging ,Cohort Studies ,Machine Learning ,03 medical and health sciences ,0302 clinical medicine ,medicine ,Humans ,Radiology, Nuclear Medicine and imaging ,Aged ,Neoplasm Staging ,Proportional Hazards Models ,Retrospective Studies ,Aged, 80 and over ,business.industry ,Soft tissue sarcoma ,Sarcoma ,Retrospective cohort study ,Middle Aged ,Prognosis ,medicine.disease ,Precision medicine ,Neoadjuvant Therapy ,Biomarker (cell) ,Random forest ,Survival Rate ,Radiation therapy ,Oncology ,030220 oncology & carcinogenesis ,Disease Progression ,Prognostic model ,Female ,Artificial intelligence ,Risk assessment ,business ,computer - Abstract
Current prognostic models for soft tissue sarcoma (STS) patients are solely based on staging information. Treatment-related data have not been included to date. Including such information, however, could help to improve these models. A single-center retrospective cohort of 136 STS patients treated with radiotherapy (RT) was analyzed for patients’ characteristics, staging information, and treatment-related data. Therapeutic imaging studies and pathology reports of neoadjuvantly treated patients were analyzed for signs of response. Random forest machine learning-based models were used to predict patients’ death and disease progression at 2 years. Pre-treatment and treatment models were compared. The prognostic models achieved high performances. Using treatment features improved the overall performance for all three classification types: prediction of death, and of local and systemic progression (area under the receiver operatoring characteristic curve (AUC) of 0.87, 0.88, and 0.84, respectively). Overall, RT-related features, such as the planning target volume and total dose, had preeminent importance for prognostic performance. Therapy response features were selected for prediction of disease progression. A machine learning-based prognostic model combining known prognostic factors with treatment- and response-related information showed high accuracy for individualized risk assessment. This model could be used for adjustments of follow-up procedures.
- Published
- 2018