Quantifying the information in noisy epidemic curves

Parag, Kris V; Donnelly, Christl A.; Zarebski, Alexander E.

doi:10.1038/s43588-022-00313-1

Cited by 17 publications

(30 citation statements)

References 64 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…(1) has successfully been applied to model many diseases including COVID-19, Ebola virus disease, pandemic influenza and measles, among others, it has one major flaw – it assumes that the generation time distribution is fixed or stationary and known [9]. If this assumption holds (we ignore surveillance biases [9,23] until the Discussion), Eq. (1) allows epidemic transmissibility to be summarised in fluctuations of the time-varying R t parameters.…”

Section: Resultsmentioning

confidence: 99%

“…First, we only examined biases inherent to R due to the difficulty of measuring the generation time accurately and across time. While this is a major limitation of existing transmissibility metrics [15], practical surveillance data are also subject to under-reporting and delays, which can severely diminish the quality of any transmissibility estimates [23,42,44]. While W ameliorates issues due to generation time mismatch, it is as susceptible as R and r to surveillance biases and corrective algorithms (e.g., deconvolution methods [45]) should be applied before inferring W. Second, our analysis depends on renewal and compartmental epidemic models [22].…”

Section: Growth Rmentioning

confidence: 99%

See 1 more Smart Citation

Angular reproduction numbers improve estimates of transmissibility when disease generation times are misspecified or time-varying

Parag

Cowling

Lambert

2022

Preprint

Self Cite

View full text Add to dashboard Cite

We derive and introduce theangular reproduction number, Ω, which measures time-varying changes in epidemic transmissibility resulting from variations in both the effective reproduction number,R, and the generation time distribution,w. Predominant approaches for tracking the dynamics of pathogen spread either inferRor the epidemic growth rater. However,Ris easily biased by mismatches between the assumed and truew, whileris difficult to interpret in terms of the individual-level branching process underpinning transmission. Moreover,Randrmay disagree on the relative transmissibility of two epidemics or variants (i.e.,rA>rBdoes not implyRA>RBfor variants A and B). We find that Ω responds meaningfully to mismatches inwwhile maintaining most of the interpretability ofR. Additionally, we prove that Ω > 1 if and only ifR> 1 and that Ω agrees withron the relative transmissibility of pathogens. Estimating Ω is no harder than inferringR, uses existing software, and requires no generation time measurement. These advantages come at the expense of selecting one free parameter. We propose Ω as a useful statistic for tracking and comparing the spread of infectious diseases that may better reflect the impact of interventions when those interventions concurrently change bothRandwor alter the relative risk of co-circulating pathogens.

show abstract

Section: Resultsmentioning

confidence: 99%

Section: Growth Rmentioning

confidence: 99%

Angular reproduction numbers improve estimates of transmissibility when disease generation times are misspecified or time-varying

Parag

Cowling

Lambert

2022

Preprint

Self Cite

View full text Add to dashboard Cite

show abstract

“…Since the α j are design variables subject to the conservation constraint in Eq. (4) , we can leverage experimental design theory [20,29] to derive novel consensus statistics to replace the default formulation from Eq. (2) .…”

Section: Resultsmentioning

confidence: 99%

“…If p = 2, this uncertainty can be circumscribed by an ellipse in the space spanned by R 1 and R 2 , and designs have a geometric interpretation as in Figure 1 . A -optimal designs minimise the bounding box of the ellipse, while D and E -optimal designs minimise its area (or volume, when extending to higher dimensions) and largest chord respectively [29,30]. These designs yield optimal versions of Λ j , , computed as shown in Eq.…”

Section: Resultsmentioning

confidence: 99%

Risk averse reproduction numbers improve resurgence detection

Parag

Obolski

2022

Preprint

Self Cite

View full text Add to dashboard Cite

The effective reproduction number R is a prominent statistic for inferring the transmissibility of infectious diseases and the effectiveness of interventions. R purportedly provides an easy-to-interpret threshold for deducing if an epidemic will grow (R > 1 ) or decline (R < 1 ). We posit that this interpretation can be misleading and overconfident when applied to infections aggregated from groups with heterogeneous dynamics. In these settings, R implicitly weights the dynamics of groups by their number of circulating infections. This induces losses in sensitivity to the dynamics of high-risk groups that may spur resurgences, promotes premature indications of epidemic control, and renders the R = 1 threshold uninformative. Applying E-optimal experimental design theory, we derive the risk averse reproduction number E to ameliorate these issues. Using analytic approaches, simulations, and a real-world case study, we find that E achieves more meaningful consensus of the dynamics across groups without being overconfident in its estimates. An E>1 generates timely resurgence signals (upweighting the impact of high-risk groups), while E < 1 ensures constituent sub-outbreaks are under control. We propose E as an alternative to R for informing policy and assessing transmissibility over large scales (e.g., state or nationwide), where well-mixed assumptions break down.

show abstract

“…However, it is difficult to choose among several data streams when each is subject to its own trade-offs. As reported in Nature Computational Science, Kris Parag and colleagues use analytic calculations informed by information theory to derive a calculable, interpretable and pathogenagnostic metric that quantifies how much information different data streams convey about real-time transmission during an epidemic 1 .…”

mentioning

confidence: 99%

Getting the most out of noisy surveillance data

McGough

2022

Nat Comput Sci

View full text Add to dashboard Cite

Quantifying the information in noisy epidemic curves

Cited by 17 publications

References 64 publications

Angular reproduction numbers improve estimates of transmissibility when disease generation times are misspecified or time-varying

Angular reproduction numbers improve estimates of transmissibility when disease generation times are misspecified or time-varying

Risk averse reproduction numbers improve resurgence detection

Getting the most out of noisy surveillance data

Contact Info

Product

Resources

About