The HBV PRoGReSs model is better than before

The PRoGreSs is a Markov model that forecasts the progression of HBV infected individuals at the national/regional level while also taking into account the impact of vaccination and horizontal transmission. The model is named after its designers: Ken Pasini, Homie Razavi, Ivane Gamkrelidze, and Devin Razavi-Shearer. It has undergone 30+ revisions/updates since its inception over two years ago, to the dismay of the epidemiology team that has to calibrate the model for 100+ countries. The latest revision has resulted in the most accurate predictive model yet. There are a number of requirements we use to test the predictability of the model:

  • Calibrate the model to the earliest available prevalence (by age) study and check to see if the model can predict the more recent prevalence (by age) studies
  • Turn off all vaccination and see if the model can predict a relatively flat prevalence by age group (for an example, see Liang 2013 study in China)
  • Check to see if the model can predict the percentage of the population who are anti-HBc positive before vaccination started

The latest model was checked against empirical data in China, US, South Korea and Iran. These were all countries that had robust prevalence studies at two points in time. In every case, the model was able to forecast the prevalence by age in the latter study. In China, the model was calibrated to HBsAg prevalence before vaccination started and then compared against the results after vaccination (see Liang 2013). In the other countries, the first and second calibration already included HBV vaccination. In all cases, there was good agreement between the model forecast and the empirical data. Please see our upcoming publication for more details.

Polaris Observatory was developed by:  
Made possible through funding from:
John Martin Foundation

Copyright © All rights reserved