Group testing is efficient to classify units as infected or not when the underlying model for the test result of units is Bernoulli distribution when the proportion of infected units in the population is small. Group testing is also preferred in estim...
Group testing is efficient to classify units as infected or not when the underlying model for the test result of units is Bernoulli distribution when the proportion of infected units in the population is small. Group testing is also preferred in estimating the population infection rate to one-by-one test procedure. In most of classical approaches to estimation, no dilution effect is assumed and the classical estimator, mle, is used. But the dilution effect happens in any area, for example, many clinical researches. The dilution effect underestimates the population infection rate and have all individuals in the group be free to go from further classification even it is infected by serious disease. In this paper, an estimation method of infection rate and the probability of being diluted is proposed when dilution effects exist and retesting scheme is feasible only once.