Accurate inference for the Youden index and its associated cutoff point based on the gamma and inverse Gaussian distributed assumption

The Youden index is often used to measure the effectiveness of biomarkers and aids to find the optimal cutoff point. Since pooled specimens have been shown to be an effective cost-cutting technique, we proposed the exact inferential procedures for the Youden index and its associated cutoff point based on the pooled specimens under the gamma or the inverse Gaussian assumption. The generalized confidence intervals (GCIs) were proposed for the Youden index and its associated cutoff point. Monte Carlo simulations were used to assess the performance of the proposed GCIs. The simulation results show that the proposed GCIs outperformed existing methods such as the bootstrap- $p$ CIs in terms of the coverage probability. Finally, the proposed procedures were illustrated by an example.