Skip to content

ardimirzaei/HISB_Predictors_Network_Analysis

Repository files navigation

Predictors of Health Information–Seeking Behavior: Systematic Literature Review and Network Analysis

Available at JMIR

Please cite as:

Mirzaei A, Aslani P, Luca EJ, Schneider CR Predictors of Health Information–Seeking Behavior: Systematic Literature Review and Network Analysis J Med Internet Res 2021;23(7):e21680 doi: 10.2196/21680

File Explanation

The main.R file contains all the steps to reproducing the work in the published paper above.

The HISB_Predictors_Coded.csv file contains the list of the significant predictors. This file is loaded to build the network analysis. It also contains the information regarding theories.

Finally, the Functions_and_Codes.R file contains the custom functions used to make the networks.

Buliding the Data Files

In main.R, the file can be downloaded straight from this github. You can aso download a local copy, and load it that way. It is important to oad the custom functions as it makes it run the code.

Once you run the function to Clean_and_Correct_Database, then you need to Produce_Edges and Produce_Nodes.

There was the ability to remove names to see connections between lower levels predictors rather than their super class (e.g. sociodemographic, health and internet). This may not work with abbreviated column names. The abbreviation list is in the custom functions and code, if it needs to be revered. Alternatively, you can pass a list of terms, and it will remove them from coding.

You can specify a cutoff (e.g. 2008) and then create the edge and node list for thpse years. Once the edge aand node list has been made, you can then export it to a CSV to use in Gephi (version 0.9.2).

Generating the Graphs

The function plot_networks prints the network. I chose to save it to a PDF but you can also plot it in the window. The graphs plo using the Fruchterman-Reingold algorithm, with some adjustments so that the labels do not overlap. The nodelist contains the membership for each nde, however, these memberships are made according to the whole grpah, rather than from the cutoff points. That is, if the whole network (made with al years combined) has 3 membership groups, then throughout the years the node colorings are produced using those 3 memberships, even though one year may have a network that has more than 3 membership groups.

Extra outputs

The last part of the main.R produces the outputs needed for the paper and the spread of the theories.

Notice any issues, please get in touch!

About

Network analysis and code for Health Information Seeking Behaviour

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages