Are you looking for new ways to use your data science skills to refine cancer treatment? In a recent study, partially funded by NCI, Drs. Ji Eun Park and Ramon Francisco Barajas Jr., built and tested a model to predict survival from glioblastoma, a particularly aggressive and devastating cancer. In the process, they learned what type of treatment yielded the best outcomes.

Using the survival model, which blends imaging and clinical data, they were able to classify patients into four distinct risk groups, based on factors such as the surgery used, patient age, as well as genetic biomarkers linked to glioblastoma treatment response.

They found that patients who had the most radical surgery (i.e., where the tumor and surrounding tissue were both removed) had the best outcomes.

Coauthor and NCI grantee, Dr. Barajas, Jr., of Oregon Health and Science University, said, “Using our model we found that an aggressive surgical approach, with maximum or complete resection of MRI abnormal areas surrounding the tumor, resulted in the most positive prognosis for patients with certain clinical features.”

He added, “We suspected that patients undergoing more aggressive surgery would have positive outcomes. However, applying our model to data from a prospective study, which included other forms of treatment and patient variables, gave us a truly objective assessment of this surgical approach and its impact on survival.”