by Yong Dam Jeong, William S. Hart, Masahiro Ishikane, Kwang Su Kim, Jong Hyuk Byun, Il Hyo Jung, Montie T. Harrison, Kazuyuki Aihara, Norio Ohmagari, Christopher B. Brooke, Ruian Ke, Robin N. Thompson, Shingo IwamiDuring the COVID-19 pandemic, rapid antigen tests (RATs) were used to detect infections, improving the effectiveness of targeted non-pharmaceutical interventions (NPIs). However, RATs based on either nasal swab or saliva samples were used, raising the question as to which type of RAT is most effective at detecting viral infections. Here, we develop a model-driven computational framework to assess different RATs and identify the most suitable test for a specified purpose, such as infection screening or determining the end of isolation, for various viral infections. Using symptomatic COVID-19 cases as a case study, we found that saliva-based RATs reduced transmission risk on average by 6.2% (95% CI: 6.1 – 6.3) compared to nasal-based RATs in the pre-symptomatic period. In addition, by ending isolation of infected individuals who have developed symptoms when consecutive RATs return negative results, the mean risk of transmission was reduced by 5.9% (95% CI: 5.7 – 6.1) using a saliva-based RAT compared to using a nasal-based RAT. These findings suggest that saliva RATs may be a useful option for mitigating SARS-CoV-2 transmission effectively. However, real-world variability in test sensitivity and sample collection should be carefully considered when evaluating the practical use of each RAT type. Our novel approach can be applied to other viruses and types of tests, enabling its use to inform public health policy decisions about which types of RAT to prioritize in future infectious disease epidemics.