To impact the COVID-19 pandemic, lifesaving antiviral therapies must be identified. The number of clinical trials that can be performed is limited. We developed mathematical models to project multiple therapeutic approaches. Our models recapitulate off-treatment viral dynamics and predict a three-phase immune response. Simulated treatment with remdesivir, selinexor, neutralizing antibodies or cellular immunotherapy demonstrates that rapid viral elimination is possible if in vivo potency is sufficiently high. Therapies dosed soon after peak viral load when symptoms develop, may decrease shedding duration and immune response intensity, but have little effect on viral area under the curve (AUC), which is driven by high early viral loads. Potent therapy dosed prior to viral peak during pre-symptomatic infection, could lower AUC. Drug resistance may emerge with a moderately potent agent dosed before viral peak. Our results support early treatment for COVID-19 if shedding duration and not AUC is most predictive of clinical severity.

SARS-CoV-2 is difficult to contain because many transmissions occur during pre-symptomatic infection. Unlike influenza, most SARS-CoV-2 infected people do not transmit while a small percentage infect large numbers of people. We designed mathematical models which link observed viral loads with epidemiologic features of each virus, including distribution of transmissions attributed to each infected person and duration between symptom onset in the transmitter and secondarily infected person. We identify that people infected with SARS-CoV-2 or influenza can be highly contagious for less than one day, congruent with peak viral load. SARS-CoV-2 super-spreader events occur when an infected person is shedding at a very high viral load and has a high number of exposed contacts. The higher predisposition of SARS-CoV-2 towards super-spreading events cannot be attributed to additional weeks of shedding relative to influenza. Rather, a person infected with SARS-CoV-2 exposes more people within equivalent physical contact networks, likely due to aerosolization.