Improved SIR - Covid-19 Prediction Model

SIR (Susceptible-Infected-Removed) model to estimate the peak while a data-driven approach based on past outbreaks is used to predict the decline of the epidemic.

Forecast Model

We employ an improved SIR model to make forecasts. The classical SIR (Susceptible-Infected-Removed) model to estimate the peak while a data-driven approach based on past outbreaks is used to predict the decline of the epidemic. The SIR model requires only two parameters- infection rate and recovery rate- to predict the progression of the epidemic. These parameters are estimated based on the best fit with the infection and recovery data. This simple model in many ways is more reliable than more complex models in which some underlying parameters are assumed in absence of reliable data. Our model was one of the earliest to predict the peak date and caseload of second wave in India (https://www.medrxiv.org/content/10.1101/2021.04.17.21255665v1).

Other parameters characterizing the epidemic:

Effective Reproduction Number (Rt)

The effective reproduction number (Rt) is roughly a ratio of infection and recovery rates that describes real-time transmission dynamics for an ongoing epidemic. Rt > 1 suggests that the infection rate is higher than the recovery rate, therefore the epidemic is growing. Similarly, Rt < 1 indicates that the epidemic is shrinking.

Test Positivity Rate (TPR)

TPR indicates total number of infected cases out of total tests conducted. This parameters, therefore, characterizes the spread of epidemic (higher the TPR, more the spread). We employ daily data based on seven day average for estimation of TPR. WHO recommends TPR to be less than 5% to effectively control the epidemic.

Case Fatality Rate (CFR)

CFR is the ratio of total number of deceased cases and total number of infected cases. This parameters, therefore, characterizes the fatality of virus as well as the preparedness of the region in terms of healthcare infrastructure.

Dataset

All the parameters estimation and predictions are based on dataset available at https://www.covid19india.org/ and https://ourworldindata.org/.

References:

  • Rajesh Ranjan, Aryan Sharma, and Mahendra K. Verma, Characterization of the Second Wave of COVID-19 in India, (2021). medrxiv, Under production in “Current Science” journal
  • Rajesh Ranjan. Temporal Dynamics of COVID-19 Outbreak and Future Projections: A Data-driven ApproachTransactions of Indian National Academy of Engineering (2020), pp.1-7. DOI:1007/s41403-020-00112-y