How disease risk awareness modulates transmission: coupling infectious disease models with behavioral dynamics

Code for the preprint https://t.co/cwj59tO9XP?amp=1

𝗧𝗶𝘁𝗹𝗲: How disease risk awareness modulates transmission: coupling infectious disease models with behavioral dynamics.

𝗔𝘂𝘁𝗵𝗼𝗿𝘀: Jaime Cascante-Vega, Samuel Torres-Florez, Juan Cordovez, Mauricio Santos-Vega

𝗔ffiliation: Grupo de Biología y Matemática Computacional (Universidad de los Andes, Bogotá D.C. Colombia)

Dependencies

python 3.7
pandas
numpy
scipy
networkx
python-louvian 

Setup

1. Make sure to have dependencies installed
2. Clone our repository https://github.com/biomac-lab/Epidemiology_behavior_dynamics.git
3. Create networks
4. Simulate (may take a lot of time!)
5. Generate figures

Usage

Creation of networks using networkx

In this investigation, we account for three different types of networks where we ran simulations on: Scale-free, Watts-Strogatz small world and grid topologies. To create these networks that are used in simulation, you need to define the number of nodes (n) with --num_nodes <n>. If not, it is set to n=1000 automatically. Hence, if you want to create all the networks specify --all True. To do this, run:

python models/create_networkxs.py --num_nodes 1000 --all True 

Conversely, if you want to create a specific network (say scale_free, small_world or grid ) with a given number of nodes (n), set --specific_network <network_name>. To create a scale-free network with 5000 nodes, run:

python models/create_networkxs.py --num_nodes 5000 --specific_network scale_free

Run simulations

In order to run a simulation, a combination of the infection probability (beta) and awareness (sigma) must be given. This combination are specified on the files beta_search.csv and sigma_search.csv located on the run/param_search folder. Each file will contain a row with the parameter value and a key for saving the result (as shown below). We test the model on an interval between 0-1, this means the files <>_search.csv contains values in the range of 0 and 1 with space intervals of 0.02 as decribed below:

key,value
000,0.00,
002,0.02,
.
.
098,0.98,
100,1.00,

The default setting for sigma_search.csv are:

key,value
100,1.00
070,0.70
050,0.50

And for beta_search.csv are:

key,value
060,0.60
070,0.70

Besides, the other entries to the model are the network_type, the network_name (output name after creating the network(s)), the type_sim which may be specified as global or local, the number of iterations n_iters (20 by default), and the length of simulation given by max_time which is set to 150 days. If you want to change the number of iterations (iters), add --n_iters <iters> to the command. To change the length of simulation (days) add --max_time <days>. The execution is then: python run/run_sims.py --network_type <> --network_name <> --num_nodes <> --type_sim <>.

For running simulations over a scale-free network with 1000 nodes in both information transmission scheme (as shown in the paper)

python run/run_sims.py --network_type scale_free --network_name scale_free_1000 --num_nodes 1000 --type_sim local
python run/run_sims.py --network_type scale_free --network_name scale_free_1000 --num_nodes 1000 --type_sim global

For running simulations over a scale-free network with 5000 nodes in both infomation transmission scheme (as shown in the paper)

python run/run_sims.py --network_type scale_free --network_name scale_free_5000 --num_nodes 5000 --type_sim local
python run/run_sims.py --network_type scale_free --network_name scale_free_5000 --num_nodes 5000 --type_sim global

For running simulations over a small-world and grid network with 1000 nodes in both infomation transmission scheme (as shown in the paper)

python run/run_sims.py --network_type small_work --network_name small_work_1000 --num_nodes 1000 --type_sim local
python run/run_sims.py --network_type small_work --network_name small_work_1000 --num_nodes 1000 --type_sim global
python run/run_sims.py --network_type grid --network_name grid_1000 --num_nodes 1000 --type_sim local
python run/run_sims.py --network_type grid --network_name grid_1000 --num_nodes 1000 --type_sim global

For running simulations over a ODE (as shown in the paper)

python run/run_ode.py

Simulation for cluster analysis

In order to analyse how the community structures (i.e. clusters or hubs) affected infection and behavior, several initial conditions are tested. To generate this initial conditions for a network with 5000 individuals, run:

python run/init_conditions/create_init_conditions.py --num_nodes 5000

Each iteration of the simulation records the state of each invididual node and is saved in a .txt file. You will need to specify network_type, network_name and the type_sim as mentioned above. The infection probability beta and awareness sigma are also required. The number of checkpoitns to be saved is 8, however you can change it by addind --num_checkpoints <num> to the command. If you want to change the number of iterrations (iters), add --n_iters <iters> to the command. To change the length of simulation (days) add --max_time <days>. The execution is then: python run/init_conditions/run_checkpoints.py --network_type <> --network_name <> --num_nodes <> --type_sim <> --beta <> --sigma <>. Remember that the number of nodes needs to be specified in the config.csv file.

For for saving the checkpoint of the simulations executed over a scale-free network with 5000 nodes in both infomation transmission scheme (as shown in the paper)

python run/run_checkpoints.py --network_type scale_free --network_name scale_free_5000 --num_nodes 5000 --type_sim local --beta 6.0 --sigma 1.0
python run/run_checkpoints.py --network_type scale_free --network_name scale_free_5000 --num_nodes 5000 --type_sim global --beta 6.0 --sigma 1.0

Figures generation

Networks visualization

In order to visualize the networks you will need to specify the network_type (i.e. scale_free, small_world or grid) and the network_name as it was saved in the /networks folder. For visualizing a (already) created scale-free network with 5000 nodes, you would run:

python plots/plot_networks.py --network_type scale_free --network_name scale_free_5000

Heatmaps visualization

Plotting a heatmap means you have runned simulations over a range of values for beta and sigma. Hence, beta_search.csv and sigma_search.csv had to be modified previously. The values that will be considered by the function for plotting are the same already specified on the /param_search files. The network_type, network_name, num_nodes, type_sim and type_hm parameters need to be specified. The later defines weather the heatmap if ploted using beta of R0 in their y axis. An example is displayed as follows:

python plots/plot_heatmaps.py --network_type scale_free --network_name scale_free_5000 --num_nodes 5000 --type_sim local --type_hm R0
python plots/plot_heatmaps.py --network_type scale_free --network_name scale_free_5000 --num_nodes 5000 --type_sim global --type_hm R0

Disease and behavior dynamics over networks

In order to plot the temporal dynamics of the simulation, there are few things to consider. Each figure is created for a specific beta and a set of sigmas. These values need to be specified in the beta_plot.csv and sigma_plot.csv files found in the /plots/params_plot folder. Thee defauld values are the ones used in the paper figures. We recommend to plot only 3 values of sigma per beta for best visualization, and only call the function for a set of parameters (leaving only one value of beta in beta_plot.csv). The network_type, network_name, num_nodes, and type_hm need to be specified. For visualizing the plot presented in our paper, run:

python plots/plot_dynamics.py --network_type scale_free --network_name scale_free_5000 --num_nodes 5000 --type_hm R0
python plots/plot_dynamics.py --network_type scale_free --network_name scale_free_5000 --num_nodes 5000 --type_hm R0

Clustered dynamics over a scale-free network

We performe cluster analysis through ploting each cluster/community dynamics over time, we do this for the 3 biggest clusters. To achive this, one must have runned simulation for each time checkpoint (explained earlier). Here, we need to specify the network_type, network_name, num_nodes, iterations, beta_select, beta_key, sigma_select, and sigma_key. Note that here you need to specify the beta and sigma values and keys. Then, for visualizing clusters in the scale-free network with 5000 individuals we should run:

python plots/plot_cluster_dynamics.py --network_type scale_free --network_name scale_free_5000 --num_nodes 5000 --iterations 10 --beta_select 0.6 --beta_key 060 --sigma_select 1.0 --sigma_key 100