Limits...
Real-time numerical forecast of global epidemic spreading: case study of 2009 A/H1N1pdm.

Tizzoni M, Bajardi P, Poletto C, Ramasco JJ, Balcan D, Gonçalves B, Perra N, Colizza V, Vespignani A - BMC Med (2012)

Bottom Line: In addition, we studied the effect of data incompleteness on the prediction reliability.Real-time predictions of the peak timing are found to be in good agreement with the empirical data, showing strong robustness to data that may not be accessible in real time (such as pre-exposure immunity and adherence to vaccination campaigns), but that affect the predictions for the attack rates.Our results show that large-scale models can be used to provide valuable real-time forecasts of influenza spreading, but they require high-performance computing.

View Article: PubMed Central - HTML - PubMed

Affiliation: Computational Epidemiology Laboratory, Institute for Scientific Interchange, ISI, Torino, Italy.

ABSTRACT

Background: Mathematical and computational models for infectious diseases are increasingly used to support public-health decisions; however, their reliability is currently under debate. Real-time forecasts of epidemic spread using data-driven models have been hindered by the technical challenges posed by parameter estimation and validation. Data gathered for the 2009 H1N1 influenza crisis represent an unprecedented opportunity to validate real-time model predictions and define the main success criteria for different approaches.

Methods: We used the Global Epidemic and Mobility Model to generate stochastic simulations of epidemic spread worldwide, yielding (among other measures) the incidence and seeding events at a daily resolution for 3,362 subpopulations in 220 countries. Using a Monte Carlo Maximum Likelihood analysis, the model provided an estimate of the seasonal transmission potential during the early phase of the H1N1 pandemic and generated ensemble forecasts for the activity peaks in the northern hemisphere in the fall/winter wave. These results were validated against the real-life surveillance data collected in 48 countries, and their robustness assessed by focusing on 1) the peak timing of the pandemic; 2) the level of spatial resolution allowed by the model; and 3) the clinical attack rate and the effectiveness of the vaccine. In addition, we studied the effect of data incompleteness on the prediction reliability.

Results: Real-time predictions of the peak timing are found to be in good agreement with the empirical data, showing strong robustness to data that may not be accessible in real time (such as pre-exposure immunity and adherence to vaccination campaigns), but that affect the predictions for the attack rates. The timing and spatial unfolding of the pandemic are critically sensitive to the level of mobility data integrated into the model.

Conclusions: Our results show that large-scale models can be used to provide valuable real-time forecasts of influenza spreading, but they require high-performance computing. The quality of the forecast depends on the level of data integration, thus stressing the need for high-quality data in population-based models, and of progressive updates of validated available empirical knowledge to inform these models.

Show MeSH

Related in: MedlinePlus

Peak timing in India and Canada: simulations and real data. (A) Peak weeks of the epidemic activity in the baseline stochastic forecast output (SFO) (gray) for eight Indian cities, ordered by decreasing latitude from top to bottom. Right: map of India, showing the Indian population distribution and the subdivision in North, South, and Central regions. (B) Peak weeks of the epidemic activity in the baseline SFO (gray) for seven Canadian provinces, ordered eastward from top to bottom. Right: map of Canada, where the Canadian provinces under study are highlighted in red. The 95% reference ranges of the simulated peak week were obtained by analysis of 2,000 stochastic realizations of the model for three different values of the seasonal rescaling factor, αmin = 0.6, 0.65, and 0.7. The peak weeks reported by the surveillance are shown as color gradients, whose limits correspond to the time interval where an incidence of greater than 80% of the maximum incidence was observed. Both maps were made exclusively for this manuscript and are not subject to copyright.
© Copyright Policy - open-access
Related In: Results  -  Collection

License
getmorefigures.php?uid=PMC3585792&req=5

Figure 7: Peak timing in India and Canada: simulations and real data. (A) Peak weeks of the epidemic activity in the baseline stochastic forecast output (SFO) (gray) for eight Indian cities, ordered by decreasing latitude from top to bottom. Right: map of India, showing the Indian population distribution and the subdivision in North, South, and Central regions. (B) Peak weeks of the epidemic activity in the baseline SFO (gray) for seven Canadian provinces, ordered eastward from top to bottom. Right: map of Canada, where the Canadian provinces under study are highlighted in red. The 95% reference ranges of the simulated peak week were obtained by analysis of 2,000 stochastic realizations of the model for three different values of the seasonal rescaling factor, αmin = 0.6, 0.65, and 0.7. The peak weeks reported by the surveillance are shown as color gradients, whose limits correspond to the time interval where an incidence of greater than 80% of the maximum incidence was observed. Both maps were made exclusively for this manuscript and are not subject to copyright.

Mentions: India is roughly halved by the Tropic of Cancer. Based on information from the Indian surveillance system, we identified three regions in the country: northern, southern, and central India (see map in Figure 7A). Northern India belongs to the northern hemisphere, where the seasonality rescaling function modulates the reproductive number (Figure S2), whereas southern India is a tropical region, where the reproductive number is fixed to its reference value R0. Central India is crossed by the Tropic of Cancer, and therefore extends into both seasonal regions. Given this subdivision of the country into large regions, we examined in more detail the situation of eight large Indian cities for which influenza surveillance data were available. The pandemic wave peaked first in the cities in central and southern India, between August and October, whereas northern Indian cities experienced the activity peak later, in November and December (Figure 7A). Concerning the reference SFO results, the six cities in central and southern India are characterized by much wider reference ranges than those typically found for cities and countries in the northern hemisphere. This is due to the lack of seasonal forcing, which generally reduces stochastic effects and thus provides a smaller reference range for the SFO datasets. However, the timing reproduced by GLEAM simulations was able to capture the early wave observed in central and southern India, which was then followed by the later peak of activity experienced in the cities of Jaipur and Delhi, which belong to the northern hemisphere. The SFO seems to indicate that the real mobility and population data integrated into the model are sufficient to provide useful information on the timing of the pandemic within the country, although the error bars for the results covered a duration of 4 to 6 weeks. At the national level, the aggregation of the pandemic waves experienced in the different regions at different times resulted in a double peak of the total incidence curve, as reported by the Indian surveillance system. In our reference SFO set, the incidence curve of India presented a double peak in more than 90% of the stochastic realizations of the model, reproducing the same seasonal pattern observed in reality (see Figure S3 for a detailed analysis of the prediction at the national level).


Real-time numerical forecast of global epidemic spreading: case study of 2009 A/H1N1pdm.

Tizzoni M, Bajardi P, Poletto C, Ramasco JJ, Balcan D, Gonçalves B, Perra N, Colizza V, Vespignani A - BMC Med (2012)

Peak timing in India and Canada: simulations and real data. (A) Peak weeks of the epidemic activity in the baseline stochastic forecast output (SFO) (gray) for eight Indian cities, ordered by decreasing latitude from top to bottom. Right: map of India, showing the Indian population distribution and the subdivision in North, South, and Central regions. (B) Peak weeks of the epidemic activity in the baseline SFO (gray) for seven Canadian provinces, ordered eastward from top to bottom. Right: map of Canada, where the Canadian provinces under study are highlighted in red. The 95% reference ranges of the simulated peak week were obtained by analysis of 2,000 stochastic realizations of the model for three different values of the seasonal rescaling factor, αmin = 0.6, 0.65, and 0.7. The peak weeks reported by the surveillance are shown as color gradients, whose limits correspond to the time interval where an incidence of greater than 80% of the maximum incidence was observed. Both maps were made exclusively for this manuscript and are not subject to copyright.
© Copyright Policy - open-access
Related In: Results  -  Collection

License
Show All Figures
getmorefigures.php?uid=PMC3585792&req=5

Figure 7: Peak timing in India and Canada: simulations and real data. (A) Peak weeks of the epidemic activity in the baseline stochastic forecast output (SFO) (gray) for eight Indian cities, ordered by decreasing latitude from top to bottom. Right: map of India, showing the Indian population distribution and the subdivision in North, South, and Central regions. (B) Peak weeks of the epidemic activity in the baseline SFO (gray) for seven Canadian provinces, ordered eastward from top to bottom. Right: map of Canada, where the Canadian provinces under study are highlighted in red. The 95% reference ranges of the simulated peak week were obtained by analysis of 2,000 stochastic realizations of the model for three different values of the seasonal rescaling factor, αmin = 0.6, 0.65, and 0.7. The peak weeks reported by the surveillance are shown as color gradients, whose limits correspond to the time interval where an incidence of greater than 80% of the maximum incidence was observed. Both maps were made exclusively for this manuscript and are not subject to copyright.
Mentions: India is roughly halved by the Tropic of Cancer. Based on information from the Indian surveillance system, we identified three regions in the country: northern, southern, and central India (see map in Figure 7A). Northern India belongs to the northern hemisphere, where the seasonality rescaling function modulates the reproductive number (Figure S2), whereas southern India is a tropical region, where the reproductive number is fixed to its reference value R0. Central India is crossed by the Tropic of Cancer, and therefore extends into both seasonal regions. Given this subdivision of the country into large regions, we examined in more detail the situation of eight large Indian cities for which influenza surveillance data were available. The pandemic wave peaked first in the cities in central and southern India, between August and October, whereas northern Indian cities experienced the activity peak later, in November and December (Figure 7A). Concerning the reference SFO results, the six cities in central and southern India are characterized by much wider reference ranges than those typically found for cities and countries in the northern hemisphere. This is due to the lack of seasonal forcing, which generally reduces stochastic effects and thus provides a smaller reference range for the SFO datasets. However, the timing reproduced by GLEAM simulations was able to capture the early wave observed in central and southern India, which was then followed by the later peak of activity experienced in the cities of Jaipur and Delhi, which belong to the northern hemisphere. The SFO seems to indicate that the real mobility and population data integrated into the model are sufficient to provide useful information on the timing of the pandemic within the country, although the error bars for the results covered a duration of 4 to 6 weeks. At the national level, the aggregation of the pandemic waves experienced in the different regions at different times resulted in a double peak of the total incidence curve, as reported by the Indian surveillance system. In our reference SFO set, the incidence curve of India presented a double peak in more than 90% of the stochastic realizations of the model, reproducing the same seasonal pattern observed in reality (see Figure S3 for a detailed analysis of the prediction at the national level).

Bottom Line: In addition, we studied the effect of data incompleteness on the prediction reliability.Real-time predictions of the peak timing are found to be in good agreement with the empirical data, showing strong robustness to data that may not be accessible in real time (such as pre-exposure immunity and adherence to vaccination campaigns), but that affect the predictions for the attack rates.Our results show that large-scale models can be used to provide valuable real-time forecasts of influenza spreading, but they require high-performance computing.

View Article: PubMed Central - HTML - PubMed

Affiliation: Computational Epidemiology Laboratory, Institute for Scientific Interchange, ISI, Torino, Italy.

ABSTRACT

Background: Mathematical and computational models for infectious diseases are increasingly used to support public-health decisions; however, their reliability is currently under debate. Real-time forecasts of epidemic spread using data-driven models have been hindered by the technical challenges posed by parameter estimation and validation. Data gathered for the 2009 H1N1 influenza crisis represent an unprecedented opportunity to validate real-time model predictions and define the main success criteria for different approaches.

Methods: We used the Global Epidemic and Mobility Model to generate stochastic simulations of epidemic spread worldwide, yielding (among other measures) the incidence and seeding events at a daily resolution for 3,362 subpopulations in 220 countries. Using a Monte Carlo Maximum Likelihood analysis, the model provided an estimate of the seasonal transmission potential during the early phase of the H1N1 pandemic and generated ensemble forecasts for the activity peaks in the northern hemisphere in the fall/winter wave. These results were validated against the real-life surveillance data collected in 48 countries, and their robustness assessed by focusing on 1) the peak timing of the pandemic; 2) the level of spatial resolution allowed by the model; and 3) the clinical attack rate and the effectiveness of the vaccine. In addition, we studied the effect of data incompleteness on the prediction reliability.

Results: Real-time predictions of the peak timing are found to be in good agreement with the empirical data, showing strong robustness to data that may not be accessible in real time (such as pre-exposure immunity and adherence to vaccination campaigns), but that affect the predictions for the attack rates. The timing and spatial unfolding of the pandemic are critically sensitive to the level of mobility data integrated into the model.

Conclusions: Our results show that large-scale models can be used to provide valuable real-time forecasts of influenza spreading, but they require high-performance computing. The quality of the forecast depends on the level of data integration, thus stressing the need for high-quality data in population-based models, and of progressive updates of validated available empirical knowledge to inform these models.

Show MeSH
Related in: MedlinePlus