Maternal/Newborn health cohort model

[joehcollins]

This is the draft PR for development of the MNH cohort model for upcoming analyses/calibration

[joehcollins]

@tbhallett @matt-graham @tamuri - I'm trying to set up some functionality for the model to be able to run with a cohort of only pregnant women. I've generated a dataframe using a 100k full model run of the properties of all women at pregnancy onset for the simulation. I am using the properties of the women who became pregnant in 2024 to overlap the properties of those within the simulation at baseline.

I've managed to do this in a hacky way (see mnh_cohort_module.py) which works in practice but there are a couple of issues:

1. This only works when I put the logic in initialise_population and ensure the cohort module runs first as there are a number of modules who schedule HSIs etc off the initial population.props within initialise_simulation - theres probably a way around this with dependencies/or something cleverer (see below)?
2. A more pressing issue - the new class PatientDetails in population.py is retaining the old version of the dataframe after the dataframe has been updated with the details of the pregnancy cohort. This is causing issues with generic HSIs being scheduled based on details which are different from this class to the real dataframe. If you run test_mnh_cohort.py on this branch you'll see what I mean r.e. RTI when debugging.

When you have some time coming up I would love to discuss an alternative approach than the one i'm using now which is maybe neater/cleaner and could avoid some of these issues. I'll keep thinking in the mean time!

[joehcollins]

(@mnjowe @thewati FYI)

[mnjowe]

Thanks Joe

…

On Thu, Jun 27, 2024 at 1:20 PM joehcollins ***@***.***> wrote: ***@***.*** <https://github.com/mnjowe> @thewati <https://github.com/thewati> FYI) — Reply to this email directly, view it on GitHub <#1320 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AHXJ7BT6JQXKBV4AO5D3JF3ZJPYOTAVCNFSM6AAAAABGSL3MNWVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDCOJUGQZDEMJTHA> . You are receiving this because you were mentioned.Message ID: ***@***.***>

[joehcollins]

@matt-graham @tamuri - Tim mentioned you were thinking more on how best to approach this? Would be great to catch up to see if I can help at all