Layman’s abstract for Canadian Journal of Statistics paper on evaluation of competing risks prediction models using polytomous discrimination index

Each week, we will be publishing layman’s abstracts of new articles from our prestigious portfolio of journals in statistics. The aim is to highlight the latest research to a broader audience in an accessible format.

The article featured today is from Canadian Journal of Statistics, with the full article now available to read in Early View here.

Ding, M., Ning, J. and Li, R. (2020), Evaluation of competing risks prediction models using polytomous discrimination index. Can J Statistics.

Competing risks data arise when patients are subject to different types of failure events. For such data, it is often important to predict a patient’s outcome status at a clinically meaningful time point. For example, in a study of monoclonal gammopathy of undetermined significance (MGUS), the two competing risks events of interest are progression to a plasma cell malignancy (PCM) and death without PCM. The two events correspond to different mechanisms of disease progression. Of interest is to predict the patient’s 10-year outcome, where the outcome categories are PCM, death without PCM, and event free. Prediction can be conducted by adopting a regression model, such as a Fine and Gray model, that relates the cumulative incidence probabilities of competing risks data to a set of covariates. In this work, the authors propose an estimator of the Polytomous Discrimination Index applicable to competing risks data, which can quantify a prognostic model’s ability to discriminate among subjects from different outcome groups. The proposed methods are robust in that they allow the regression model used for prediction to be subject to potential model misspecification. Authors also develop a computationally efficient algorithm to allow easy implementation. The proposed methods feature desirable theoretical properties and performed satisfactorily in simulation studies. The practical utility of the proposed methods were demonstrated to the motivating example on MGUS.


More Details