Hongkonginfluensan i sverige
Distributions of residuals were centered at 0, indicating no systematic biases of the model S6 Fig. While R 0 followed the same seasonal cycle as years with epidemics Fig 3A , the estimated infection rates remained at near-zero over these extended time-periods when A H1N1 was not in circulation Fig 2A. In comparison, for influenza B, major epidemics tended to occur in winter months Fig 1D. In an exploratory analysis, we coupled a commonly used multi-strain susceptible-infected-recovered-susceptible SIRS model [ 24 ] with a particle filter [ 34 — 36 ], a Bayesian inference method, and applied it to the Hong Kong dataset.
For the second challenge, we combine the IF2 iterated filtering algorithm [ 41 ] with space re-probing [ 36 , 43 ], a technique that allows continuous exploring of state space. Influenza viruses undergo continual genetic mutations and punctuated antigenic changes, which allow them to escape prior immunity and cause new epidemics [ 3 — 6 ]. Influenza epidemics cause substantial morbidity and mortality every year worldwide.
Finally, it is challenging to estimate a large number of parameters needed to characterize the interactions between virus-pairs. Thick black lines show the mean excluding for the A H1N1 pandemic. The authors declare that they have no other potential competing interests. Influenza epidemics recur annually in many regions of the world and cause significant morbidity and mortality, leading to 3 to 5 million severe infections and , to , deaths worldwide each year [ 1 , 2 ].
We therefore developed a model-inference system that addresses the above challenges. Our results also suggest that A H3N2 may undergo antigenic mutations in both summers and winters and thus monitoring the virus in both seasons may be important for vaccine development. However, unlike the winter-time epidemics in temperate regions that have been well-characterized bygd seasonal forcing of climate variables humidity alone or with temperature [ 37 , 38 ], to date no models capable of capturing the diverse seasonal pattern in tropical and subtropical climates exist.
The combined SR-IF2 inference method coupled with the empirical seasonal multi-strain SIRS model is thus able to preserve the long-term epidemic feature while adeptly explore the state space to capture the epidemic dynamics.
Hongkonginfluensan - vad kan man förvänta sig av ett virus och hur man klarar det?
Second, to estimate epidemiological parameters such as the duration of immunity, the inference system must be able to preserve such long-term epidemic features. The funders had no role in study design, uppgifter collection and analysis, decision to publish, or preparation of the manuscript. The basic fortplantnings- number R 0 is a key epidemiological parameter describing the transmissibility of an infection, defined as the average number of secondary infections that arise from a primary infection in a naïve population.
These highly diverse epidemic dynamics, however, also present major challenges for inference on transmission dynamics. First, the seasonality of influenza in part shapes the epidemic dynamics and thus should be accounted for in the model.
Farliga smittor beneath 100 år
Our model-inference system only requires sub typed incidence data, commonly collected in routine surveillance, and thus could be adapted to study influenza transmission worldwide. Overall, our study reveals intricate epidemiological interactions and underscores the importance of simultaneous monitoring of population immunity, incidence rates, and viral genetic and antigenic changes. Here we develop a model-inference system and apply it to estimate these characteristics from incidence data collected in Hong Kong since We also find key ecological interactions and intricate balance that allow the three influenza viruses to co-circulate.
These recurrent epidemics are a combined outcome of viral antigenic changes, interactions among co-circulating influenza viruses, transmission and host immune response. Indeed, testing using model-generated mock epidemics indicates that this model-inference system and optimization strategy fryst vatten able to recover the true model state variables and parameters S1 Text and S2 and S3 Figs. Multi-pass filters e.
We then incorporate this R 0 seasonal cycle as the prior distribution in the model-inference system to control for the seasonality observed in Hong Kong. Improving understanding of interactions among influenza viruses and the resulting impact on epidemic dynamics will thus provide insights to aid public health efforts to mitigate the burden of influenza. As a result, the estimated R 0 seasonal cycle has clear bimodal peaks for the two A subtypes Fig 1E and 1F , although to a less extent for A H1N1 ; and for type B, it has only one peak in the winter Fig 1G.
This is an open access article distributed under the terms of the Creative Commons Attribution License , which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited. Their continuous evolution and ecological interactions shape transmission dynamics and are thus crucial for devising public health interventions e. Hong Kong is a subtropical city located in Southeast Asia.
The latter cross-reactive immunity—termed cross-immunity—leads to epidemiological interactions among influenza viruses that, along with viral antigenic changes, shape influenza phylodynamics and epidemic dynamics [ 19 — 31 ]. PLoS Comput Biol 16 6 : e Copyright: © Yang et al. Upper panel: infection rate time series for the three influenza viruses A. Color lines show infection rates for individuals years during Jan —July ; lighter colors show earlier years and darker colors for later years.
While this model-inference system was able to reproduce the observed epidemic pattern, the model state did not appear to converge see S1 Fig. Three major challenges exist. Here we apply the validated model-inference system to the Hong Kong dataset and infer how major antigenic evolution and cross-immunity affect influenza transmission dynamics therein over the last 20 years. However, it remains unclear how such ecologic competition via cross-immunity and antigenic mutations that allow immune escape impact influenza epidemic dynamics at the population level.
Utilizing long-term influenza surveillance data since , we are able to estimate the strength of cross-immunity between each virus-pairs, the timing and frequency of punctuated changes in population immunity in response to antigenic mutations in influenza viruses, and key epidemiological parameters over the gods 20 years including the pandemic. Single-pass filters e.
To estimate parameters in high dimensional state space, we take a multi-round inference strategy; in the first round, we parse the entire state space into subspaces and test each to find the optimal subspace; subsequent rounds then build on the previous one until the model estimates converge.