Improving Risk Prediction in Patients with Soft Tissue Sarcoma

Soft tissue sarcomas (STS) constitute a rare tumor entity, accounting for < 1 % of malignancies diagnosed per year. In case of localized disease, treatment mainly consists of wide resection with clear surgical margins, as well as radiotherapy. Local recurrence (LR), distant metastasis (DM), and overall survival (OS) are the main oncological outcome measures in STS patients. Available risk prediction tools incorporate established risk factors including age, tumor size, and histology. Although these models are rather rudimentary, they are still frequently used for therapy decisions.

In recent years, artificial intelligence (AI) and machine learning (ML) methods have gained importance in clinical research. Advantages of AI-based risk prediction modelling include the possibility of overcoming limitations with conventional statistical approaches used for prognostication. Furthermore, the data-driven approach of AI-based prediction tools enables analysis of patient features in an “unbiased” manner while also having the capacity to take into consideration the changing impact of variables over time, and potential variable interactions. By combining data from different European sarcoma centers, we strive to improve risk prediction in patients diagnosed with STS. The use of AI-based tools in addition to conventional statistics has the potential to help us to understand the contribution of certain factors to development of local recurrence, distant metastasis, and overall survival better.