A2076
Title: Robust estimation and efficient design for group testing under error uncertainty
Authors: Shih-Hao Huang - National Central University (Taiwan)
Chun-Ting Chen - National Central University (Taiwan) [presenting]
Zih-Jing Lin - National Central University (Taiwan)
Abstract: Group testing is widely used in public health, but its performance can be seriously affected by test errors. Adopting simplified error models can easily lead to substantial model misspecification bias, while more flexible models tend to increase the variance of the estimator. A robust estimation method is introduced that automatically downweights potentially misclassified observations to mitigate bias. An adaptive data collection strategy based on this estimation is further proposed to reduce the variance of the estimator. Simulation studies under various error mechanisms show that the proposed methods usually have competitive or even better performance relative to conventional MLE-based approaches.
