Sparse Bayesian predictive modelling of tumour response using radiomic features. Issue 1 (25th April 2022)

• Record Type: Journal Article
• Authors: Golchi, Shirin; Fu, Jingyan; Liu, Xiaoyang; Yu, Eugene; Forghani, Reza; Bhatnagar, Sahir
• Is Part Of: Stat. Volume 11:Issue 1(2022)
• Page Start: n/a
• Page End: n/a
• Publication Date: 2022-04-25
• Subjects: hierarchical model -- horseshoe prior -- missing data -- radiomics; Statistics -- Periodicals; 519.2
• DOI: 10.1002/sta4.450
• Languages: English
• ISSNs: 2049-1573

Abstract

Abstract : We propose a sparse Bayesian hierarchical model for the analysis of data including radiomic features for characterization of head and neck squamous cell carcinoma. The proposed model facilitates radiomic feature selection, handling of missing values in key predictors as well as prediction in a unified framework. The fully Bayesian approach enables adequate incorporation of uncertainty arising from various aspects of the inference and prediction procedure. The prediction performance of the model is assessed via cross validation and compared with two frequentist methods.