Abstract

Abstract Purpose: Advances in the population screening programs and treatment efficacy, as well as, the rapidly aging global population, have led to a gradual increase of the number of long-term breast cancer survivors (BCS) in the past years. With the estimation for their survival rate to continue rising, it has become evident by the clinical and scientific community that BCS should be efficiently monitored in order to obtain a better understanding of their needs and subsequently, improve their medical care, further increase their survivorship and optimize the associated healthcare costs. Methods: A novel data-mining methodology is proposed for the temporal analysis of healthcare service-use trajectories of long-term BCS. Specifically, an unsupervised clustering technique, based on Dynamic Time Warping, is applied on the extracted trajectories of a large longitudinal cohort (SURBCAN) of BCS and non-breast cancer (BC) women, with data on their contacts to all healthcare services. Introducing the time dimension permits the identification of complex time-dependent patterns (clusters) of the services that these women have used throughout the course of a five-year monitoring period. A comparison of the identified time patterns and the respective patient characteristics is also performed between the BCS (case) and non-BC (control) group. Results: More than two hundred clusters representing temporal patterns of healthcare service use, were extracted for both the case and control group. It was found that the BCS group made a more complex and intense use of certain healthcare services, such as, radiology, outpatient care and hospital admission, than the control group. Furthermore, the corresponding mortality rates were significantly higher and a larger number of comorbidities (varying between different clusters) were registered at the beginning of the monitoring period in BCS. Additional patient characteristics and time information for certain paths of the extracted service-use network were reported and analyzed. Conclusion: A flexible data-mining methodology was applied on a longitudinal cohort of long-term BCS for the identification of temporal patterns of healthcare service use, in an attempt to gain more insight into their needs, better predict their evolution, and subsequently, improve their received attention and eventually optimize the associated healthcare costs. Citation Format: Alexia Giannoula, Mercè Comas, José Martín Solorzano Gonzalez, Ferrán Sanz, Xavier Castells, Maria Sala. Identifying Temporal Patterns in Healthcare Service-Use Trajectories of Long-Term Breast Cancer Survivors [abstract]. In: Proceedings of the 11th Annual Symposium on Global Cancer Research; Closing the Research-to-Implementation Gap; 2023 Apr 4-6. Philadelphia (PA): AACR; Cancer Epidemiol Biomarkers Prev 2023;32(6_Suppl):Abstract nr 48.

Full Text
Published version (Free)

Talk to us

Join us for a 30 min session where you can share your feedback and ask us any queries you have

Schedule a call