We gratefully acknowledge support from
the Simons Foundation and member institutions.
Full-text links:

Download:

Current browse context:

stat.ME

Change to browse by:

References & Citations

Bookmark

(what is this?)
CiteULike logo BibSonomy logo Mendeley logo del.icio.us logo Digg logo Reddit logo ScienceWISE logo

Statistics > Methodology

Title: Tailored Bayes: a risk modelling framework under unequal misclassification costs

Abstract: Risk prediction models are a crucial tool in healthcare. Risk prediction models with a binary outcome (i.e., binary classification models) are often constructed using methodology which assumes the costs of different classification errors are equal. In many healthcare applications this assumption is not valid, and the differences between misclassification costs can be quite large. For instance, in a diagnostic setting, the cost of misdiagnosing a person with a life-threatening disease as healthy may be larger than the cost of misdiagnosing a healthy person as a patient. In this work, we present Tailored Bayes (TB), a novel Bayesian inference framework which "tailors" model fitting to optimise predictive performance with respect to unbalanced misclassification costs. We use simulation studies to showcase when TB is expected to outperform standard Bayesian methods in the context of logistic regression. We then apply TB to three real-world applications, a cardiac surgery, a breast cancer prognostication task and a breast cancer tumour classification task, and demonstrate the improvement in predictive performance over standard methods.
Subjects: Methodology (stat.ME); Applications (stat.AP)
Cite as: arXiv:2104.01822 [stat.ME]
  (or arXiv:2104.01822v3 [stat.ME] for this version)

Submission history

From: Solon Karapanagiotis [view email]
[v1] Mon, 5 Apr 2021 09:37:07 GMT (1093kb,D)
[v2] Tue, 17 Aug 2021 10:54:51 GMT (1094kb,D)
[v3] Fri, 15 Oct 2021 09:52:28 GMT (1909kb,D)

Link back to: arXiv, form interface, contact.