Optimal data-driven policies for disease screening under noisy biomarker measurement

Abstract

Biomarker testing, where a biochemical marker is used to predict the presence or absence of a disease in a subject, is an essential tool in public health screening. For many diseases, related biomarkers may have a wide range of concentration among subjects, particularly among the disease positive subjects. Furthermore, biomarker levels may fluctuate based on external or subject-specific factors. These sources of variability can increase the likelihood of subject misclassification based on a biomarker test. We study the minimization of the subject misclassification cost for public health screening of non-infectious diseases, considering regret and expectation-based objectives, and derive various key structural properties of optimal screening policies. Our case study of newborn screening for cystic fibrosis, based on real data from North Carolina, indicates that substantial reductions in classification errors can be achieved through the use of the proposed optimization-based models over current practices.