SU-F-T-98: Knowledge Modeling for The Outcome of Brain Stereotactic Radiosurgery

Authors


Abstract

Purpose:

To create a model that will better predict the survival time for patients that were treated with stereotactic radiosurgery for brain metastases using a support vector machine regression model.

Methods:

This study utilized data from 481 patients, which were equally divided into training and validation datasets randomly. The predictor variables for the SVM model consisted of the actual survival time of the patient, the number of brain metastases, the GPA and KPS scores, prescription dose, and the size of the largest PTV. The resulting survival time predictions were analyzed against the actual survival times by single parameter classification and two-parameter classification. The predicted mean survival times within each classification are compared with the actual values to obtain the confidence interval associated with the model's predictions.

Results:

The number of metastases and KPS scores, were consistently shown to be the strongest predictors after single parameter classification, and were thus chosen for as first classifiers for the two-parameter classification. When the survival times were analyzed with the number of metastases as the first classifier, the best correlation was obtained for patients with 3 metastases, while patients with 4 or 5 metastases had significantly worse results. When the KPS score was used as the first classifier, patients with a KPS score of 60 and 70 had similar strong correlation results.

Conclusion:

The number of metastases and the KPS score both showed to be good predictors of patient survival time. The model was less accurate for patients with more metastases and certain KPS score ranges due to the lack of training data.

Ancillary