Quantifying the information in noisy epidemic curves

Reliably estimating the dynamics of transmissible diseases from noisy surveillance data is an enduring problem in modern epidemiology. Key parameters are often inferred incident time series, with aim informing policy-makers on growth rate outbreaks or testing hypotheses about effectiveness public health interventions. However, reliability these inferences depends critically reporting errors and latencies innate to series. Here, we develop analytical framework quantify uncertainty induced by under-reporting delays infections, as well a metric for ranking informativeness. We apply this two primary sources inferring instantaneous reproduction number: epidemic case death curves. find that assumption curves more reliable, commonly made acute infectious such COVID-19 influenza, not obvious possibly untrue many settings. Our clarifies quantifies how actionable information pathogen transmissibility lost due limitations. A measuring noise different outbreak limits estimates spread developed used show series rarely better than transmissibility.