Additional data

Estimates of R(t) can also be improved by reconstructing missing data, adding delay distributions, and adding additional data streams.

Reconstruction of missing data

Section in progress

Delay distributions

Section in progress

Importantly, the definition of R(t) is linked to the data that are being used, so models that calculate a similar quantity as R(t) but instead from infections, symptom onset, or reports are important quantities but differ in definition from the instantaneous reproduction number R(t) as defined throughout the literature.

Several packages have been created to extend EpiEstim to use delay distributions:

• bayEStim: Our method extends that of Cori et al (2013), adding Bayesian imputation of missing symptom onset dates, imputation of infection times using an external estimate of the incubation period, and an adjustment for reporting delay.
• estimateR involves combining various delay distributions with EpiEstim
• EpiInvert also has methods for including delay disributions with EpiEstim

Additional data streams

Section in progress

Additional data include: clinical data, wastewater