Using Temporal Mining to Examine the Development of Lymphedema in Breast Cancer Survivors

Abstract

Secondary lymphedema is a lifetime risk for breast cancer survivors and can severely affect quality of life. Early detection and treatment are crucial for successful lymphedema management. Limb volume measurements can be utilized not only to diagnose lymphedema but also to track progression of limb volume changes before lymphedema, which has the potential to provide insight into the development of this condition. This study aims to identify commonly occurring patterns in limb volume changes in breast cancer survivors before the development of lymphedema and to determine if there were differences in these patterns between certain patient subgroups. Furthermore, pattern differences were studied between patients who developed lymphedema quickly and those whose onset was delayed. A temporal data mining technique was used to identify and compare common patterns in limb volume measurements in patient subgroups of study participants (n = 232). Patterns were filtered initially by support and confidence values, and then t tests were used to determine statistical significance of the remaining patterns. Higher body mass index and the presence of postoperative swelling are supported as risk factors for lymphedema. In addition, a difference in trajectory to the lymphedema state was observed. The results have potential to guide clinical guidelines for assessment of latent and early-onset lymphedema.