- The ABM is written in C++ and compiled with g++ in the Linux OS. All the .h and .cpp files reside in the main directory. There is also a makefile to help facilitate building. The makefile will build main(.exe). This file will use various input files that are stored in ./defaults/input/xxxx sub-directories. Check the Input-Files.docx file to get a list of the input files and their uses.
- In order to run the ABM out of main menu, you need to execute the main executable. This is done by running ./main arg1 arg2 keyword in a Linux environment. This will put output in the ./stats/scenario/job_no. directory for each scenario. You will need a copy of the ABM program (main.exe), the ./defaults/input sub-directory and an empty ./stats sub-directory. Make sure the directory that contains the executable (main.exe) also contains the ./stats and ./defaults sub-directories. Keywords control the output of different statistical outputs to the ./stats/scenario/job no. directory. Including the desired keyword will output the appropriate statistical file.
Example - ./main 8 9 10 keywords
Argument 1 = job_no. This is the sub-directory in the ./stats/scenario sub-directory where the output will be placed
Argument 2-4 = scenario numbers to run –the scenarios in the following table are the defaults used in the model. They can be changed by changing the input files in ./defaults/input/input-file-names/input-file-names.csv file. The scenarios listed (8, 9, 10, 22 and 23) are the current valid scenarios for this model.
| Scenario no. | Xpert tiers | smear tiers | Clinical diagnosis tiers | Notes |
|---|---|---|---|---|
| 8 | 1, 2, 3, 4 | none | All | District level Xpert decentralization scenario, simulated over the entire period: 2015-2030 |
| 9 | 1, 2, 3 and some 4 | none | All | Baseline Xpert placement at 170 facilities in Ghana, simulated over the entire period: 2015-2030 |
| 10 | All | none | All | Full Xpert decentralization scenario, simulated over the entire period: 2015-2030 |
| 22 | Start 9 switch to 8 | none | All | Running baseline model to 2023 and switching to district level decentralization in 2024 through 2030 |
| 23 | Start 9 switch to 10 | none | All | Running baseline model to 2023 and switching to Full decentralization in 2024 through 2030 |
Keywords = SampleStats, ResourceStats, VisualizationStats, TableauStats, PatientStats, PatientSamplingStats. The different keywords will output different statistics from the model.
The ABM will create output files: AgentStats.csv, AgentStats-Region.csv, AgentStats-District.csv and Input parameters. AgentStatsXXXX.csv files hold the year by year statistical totals for each disease, broken down at the national, regional and district results. The file InputParameters.csv will also be created that contains the randomized parameter values used for this job. The ABM will copy all the input files used for this run from ./defaults/input to a ./stats/scenario/job_no/input sub-directory.
In order to run statistic compilation after multiple ABM runs, the user will need to run stat_comp.exe executable. It runs out of ./stat-compilation sub-directory. It can be built using the makefile and .h and .cpp files in the ./stat-compilation sub-directory. It assumes that the output files for each scenario exist in ./stats/scenario sub-directory. It will store the aggregate and individual output in ./stats/scenario directory.
Example - Stat_comp 1 2 3
argument 1 = number of jobs to compile argument 2 = number of years in each job argument 3 = scenario – number of scenario that run using the model runs. Scenario numbers between 1 and 24. Same as #2 above.
The routine will create files AgentStats-Aggregate.csv and AgentStats-Aggregate-_MC.csv, AgentStats-Region-Aggregate.csv and AgentStats-Region-Aggregate-_MC.csv, AgentStats-District-Aggregate.csv and AgentStats-District-Aggregate-_MC.csv. The first file will compile averages for all the statistical values. The second file contains output from each of the runs for Monte Carlo analysis.