Pastells / covid_models

Collection of epidemic models. Includes ODEs, stochastic and complex networks, with non-markovian distributions (Erlang type).

Epidemic Models

Epidemic models in python to be callibrated to describe the Covid-19 pandemic. These models were used to test EpidemicGGA.

• Paper (open access) - ePyDGGA: automatic configuration for fitting epidemic curves
• Documentation: https://ulog.udl.cat/static/doc/epidemic-gga/html/index.html

Algorithms used:

• ODEs - deterministic continuous mean-field
• Discrete - deterministic discrete mean-field
• Gillespie - Stochastic mean-field
• Event-driven algorithm (fast-sir variant)
• Event-driven algorithm (with complex network). An ER or BA network can be used.

Stochastic models use exponential or Erlang distributed times (Erlang models).

Implemented models:

• SIR
  • ODEs (Erlang) (with sections)
  • Gillespie (Erlang) (with sections)
  • Fast (with sections)
  • Network (with sections)
• SIRD
  • Discrete ODEs
  • Gillespie
  • Gillespie parallelized version
• SAIR
  • Gillespie (Erlang) (with sections)
  • ODEs (with sections)
  • Fast (with sections)
  • Network (with sections)
• SEAIR
  • Gillespie

Additionally, the following models have been adapted from the literature:

• SIDARTHE model (Matlab)
• SIDARTHE with sections (Python)
• MMCAcovid19 (Julia)
• SEIPAHRF

For the mean-field ones: beta, beta_a, delta, delta_a, alpha and n are all changed continuously using a tanh, for the network ones n changes abruptly.

How to run the models

To list all the available models, use:

python -m models --help

To execute a model, use:

python -m models <model> [model args]

To see the arguments for a specific model (p.e. sird), use:

python -m models sird --help

Auto-configurable models:

The following models are ready to be used with Optilog:

• SIR
  • SIR (Gillespie + ODEs)
  • SIR with sections (ODEs)
  • Network SIR
  • Network SIR with sections
  • Erlang SIR
  • Erlang SIR with sections
• SIRD
  • SIRD
• SAIR
  • SAIR
  • Network SAIR
  • Network SAIR with sections
  • Erlang SAIR
  • Erlang SAIR with sections
• SEAIR
  • SEAIR
• SIDARTHE (Original in Matlab)
• SIDARTHE with sections (Python)