Predicting Breast Cancer Mortality in the Presence of Competing Risks Using Smartphone Application Development Software
Pages 322-330
Yuanyuan Liu, Ellen P. McCarthy and Long H. Ngo
DOI: http://dx.doi.org/10.6000/1929-6029.2015.04.04.2
Published: 03 November 2015

Abstract: The widespread use of smartphone applications (apps) provides a promising new platform for medical research and healthcare decision making. Given the need to help guide clinical discussions about the appropriateness of breast cancer screening in the presence of competing risks among older women, we proposed to incorporate the Fine-Gray prediction model, which offers more intuitive clinical interpretation of risk in the presence of competing risks, into a smartphone-based decision aid application. Clinicians can input the woman’s characteristics and medical history, and the app will output prediction estimates of both types of events (i.e. death from breast cancer and competing risk events) given the presence or absence of breast cancer screening. This prototype was built using drag-and-drop visual programming tools provided by the free, cloud-based software “MIT App Inventor for Android.” It will be intended for clinicians to use in the context of patients’ values to decide whether screening is appropriate for an individual. Our analysis indicated that screening was beneficial to survival, and that older women benefited less from screening due to the increasing incidence of non-breast-cancer competing risk deaths as age increased. The algorithm we implemented for the app provides instant probability estimates that help quantify screening benefits as a function of age, and comorbidity burden.