numalariamodeling / covid-chicago Goto Github PK

-add the updated python script for the extractions to the repo

• excel files are located here \projects\covid_chicago\emod_sim\age_contact_matrices
  
• and the output of the extraction should be a csv file as in projects\covid_chicago\cms_sim\inputs contact_matrix_grp4.csv - adding the location to the csv filename
• extract for US only
  the age groups are

1. • 0-19
2. • 20-39
3. • 40-59
4. • 60 to 100

and

1. 0 to 9
2. 10 to 19
3. 20 to 29
4. 30 to 39
5. 40 to 49
6. 50 to 59
7. 60 to 69
8. 70 +

Could be a script in R or in python that reads in the scenarios.csv and the generated trajectories.csv files and counts the one that are missing - and exports it to a small txt file for easy access/review. Useful for larger simulation runs.

• since the age groups were updated and run with different age bins, age specific parameter will need to be re-checked from literature starting from parameters_default_0401.csv file in cms_sim/ inputs

• NMH catchment = catchment pop * market share
• With varying Ki
• using latest extended-cobey emodl
• check if Ki * pop = 2.5e-7 * 2700000 = Ki_adjusted *NMHcatchment
• ~KI adjusted ~ 2.9e-8

• Ki to be sampled inside the generateParameterSamples or outside are currently done

Simulations Ki sampled outside the sampling function:

• 20200405_mr_test_Cook_higherCrit_sameCFR_rn45
• 20200405_mr_test_Cook_lowerCrit_lowerCFR_rn16

Corresponding simulations Ki sampled with others:

• 20200405_mr_test_Cook_higherCrit_sameCFR_kiuniform_rn27
• 20200405_Cook_higherCrit_sameCFR_kiuniform_rn96

• adding more time-events for the detection parameter

• add parameter for varying stop time of social distancing time event
• adapt analyzer to visualize scenarios

20200406mr_NMH_catchment_v1_rn61

• peak of the epi-curve
• growth rate
• mortality rate
• ...

keep track or error messages and log of simulations that run locally

• similar to the age-model postprocessing script ?
• visualise epi curves for each EMS region

To run the CMS model a python module setup exists in "dtk-tools: CMS".
However dtk-tools-CMS has been inactive for some time and the present workflow entails some features that are not included in dtk-tools, whereas the structure of dtk-tools is much more robust and thought-through. Ideally the best features of both approaches can be combined.

age_matrix_means

• updated age groups
• add code to ensure symmetry of contacts i.e age1-2 == age2==1
• add code to rescale sum of values to 1
• extend script to extract values from the other available excel files (home, school, work same format)

Currently the code is hardcoded for a single's user setup, to add:

• run for other users using their custom path specifications in load_paths.py
• adapt the shell submission script in the runScenarios.py files
• run for the NU group members on the quest project folder (general setup)

On quest:
Each user to setup the required folder structure

currently using placeholder from the previously defined 4 age groups

• and saving input parameter into a txt file (in addition to the sampled parameter)-
• input parameter include: start date, duration, number of runs, population,...

• socialDistance_time changes for Cook or NMH catchment population and possibly for other scenarios as well

specified two times in emodl generator and in runScenarios.py

• only the one in the emodl generator is effective

• updated age groups
• ensure symmetry of contact matrix (despite some randomness in source data)
• re-scale contact values to 1

as done in the EMOD version, i.e.
contacts that happen at home, at school or at work, this will allow a finer specification of social distancing interventions, i.e. school closure (no contacts in schools but increased contact at home ?)

remove
(reaction detect_As_{} (As_{}) (As_det1_{}) (* d_As As_{}))
(reaction detect_symp_{} (Sym_{}) (Sym_det2_{}) (* d_Sym Sym_{}))
(reaction detect_hosp_{} (Sys_{}) (Sys_det3_{}) (* d_Sys Sys_{}))

• calculate mortality rate per acuity group using the format in the attached picture
  

the age model did run within seconds (15 trajectories in 6min), but the postprocessing needs debugging (again?)

suffix_names = [x.split('_')[1] for x in df.columns.values if 'susceptible' in x]

Since the column names include now two underscores should it be ?
suffix_names = [x.split('_age')[1] for x in df.columns.values if 'susceptible' in x]
But the suffix_names are empty.

Examples

• Caught exception during execution: Count is less than 0 (P:-1)

• Caught exception during execution: Count is less than 0 (E:-1)

• therefore often less samples are succesfully producing trajectories.csv than are defined in the beginning

Run age model, adjust fitting parameters to get good fit per covidregion. generate relevant age plotters and compare to covid data per age.

Specifying age groups for each EMS region and contact matrix

• currently hard coded for 4 and 6 specific age groups

for example
workflow functions

1. generate_and_run_scenarios.py
2. postprocess_and_sample_plot.py

helper functions

• load_paths.py
• parameter_helpers.py
• simulation_helpers.py
• processing_helpers.py

or look at setup as managed in dtk-tools cms

https://github.com/InstituteforDiseaseModeling/dtk-tools/tree/master/examples/CMS/CMSCalibration

• how does it work and can it be used with covid-cms emodls ?