Open mmcleod89 opened 7 months ago
Hi @UCL/hiv-modelling, here are some early epidemic comparison graphs of mean averages between 25 HIVpy runs (population size: 100k) and the 100 core SAS runs kindly provided by Jenny.
Population (15-49)
→ s_alive1549
Population (15-49, male)
→ s_alive1549_m
Population (15-49, female)
→ s_alive1549_w
Deaths (tot)
→ s_dead_all
Non-HIV deaths (tot)
→ s_dead_all
- s_death_hiv
Non-HIV deaths (ratio)
→ (s_dead_all
- s_death_hiv
) / (s_alive_m
+ s_alive_w
)
At least 1 short term partner (ratio)
→ s_newp_ge1
/ (s_alive_m
+ s_alive_w
)Short term partners (15-49, male)
→ (s_m_1524_newp
+ s_m_2534_newp
+ s_m_3544_newp
) / s_alive1549_m
Short term partners (15-49, female)
→ (s_w_1524_newp
+ s_w_2534_newp
+ s_w_3544_newp
) / s_alive1549_w
@mmcleod89 is currently in the process of making improvements to the partner balance regulation code so we don't need to take too much stock of these graphs, but having them here for later comparison won't hurt.
Partner sex balance (15-24, male)
→ log(m15r
, 10)Partner sex balance (15-24, female)
→ log(w15r
, 10)Partner sex balance (25-34, male)
→ log(m25r
, 10)Partner sex balance (25-34, female)
→ log(w25r
, 10)Partner sex balance (35-44, male)
→ log(m35r
, 10)Partner sex balance (35-44, female)
→ log(w35r
, 10)
Thanks @pineapple-cat Good to start to compare these graphically. For the earlier graphs I would suggest having the y axis start at 0 to enable comparison visually.
I'm afraid I won't be on the call this coming Tuesday but hopefully you all can go ahead and discuss them.
Hi Emily,
I’ve just noticed a couple of things, highlighted below
From: Emily Dubrovska @.> Sent: Friday, February 16, 2024 8:33 PM To: UCL/hivpy @.> Cc: Cambiano, Valentina @.>; Team mention @.> Subject: Re: [UCL/hivpy] Early Epidemic Comparisons (Issue #175)
⚠ Caution: External sender
Hi @UCL/hiv-modellinghttps://github.com/orgs/UCL/teams/hiv-modelling, here are some early epidemic comparison graphs of mean averages between 25 HIVpy runs (population size: 100k) and the 100 core SAS runs kindly provided by Jenny.
Population
Comparison.of.Population.15-49.Over.Time.png (view on web)https://github.com/UCL/hivpy/assets/34184448/15e2c5e7-d907-48b8-abff-1142d2fe4a89 Comparison.of.Population.15-49.Male.Over.Time.png (view on web)https://github.com/UCL/hivpy/assets/34184448/4c4b42d7-baf6-42ed-859a-259d0bfb4477 Comparison.of.Population.15-49.Female.Over.Time.png (view on web)https://github.com/UCL/hivpy/assets/34184448/0b16e4e3-00a8-47eb-9f3f-fcaaf9f08553
Deaths
Deaths (tot) → s_dead_all
Non-HIV deaths (tot) → s_dead_all - s_death_hiv
Non-HIV deaths (ratio) → (s_dead_all - s_death_hiv) / (s_alive_m + s_alive_w) Should this not be the total number of deaths? Or are you trying to calculate the death rate? Comparison.of.Deaths.Tot.Over.Time.png (view on web)https://github.com/UCL/hivpy/assets/34184448/c0298620-4ceb-4871-ac3c-6b72420a95fa Comparison.of.Non-Hiv.Deaths.Tot.Over.Time.png (view on web)https://github.com/UCL/hivpy/assets/34184448/92445699-233c-4672-b333-549c0fecc2ab Comparison.of.Non-Hiv.Deaths.Ratio.Over.Time.png (view on web)https://github.com/UCL/hivpy/assets/34184448/a2f2b636-1606-43ab-9097-18f5e0905f3e Short Term Partners
At least 1 short term partner (ratio) → s_newp_ge1 / (s_alive_m + s_alive_w)
Short term partners (15-49, male) → (s_m_1524_newp + s_m_2534_newp + s_m_3544_newp) / s_alive1549_m
Short term partners (15-49, female) → (s_w_1524_newp + s_w_2534_newp + s_w_3544_newp) / s_alive1549_w Note that only people 15-64 have condomless sex partner, I think s_alive_m and s_alive_w are for the same age group. It seems like you are missing s_m45_49 from the numerator
Comparison.of.At.Least.1.Short.Term.Partner.Ratio.Over.Time.png (view on web)https://github.com/UCL/hivpy/assets/34184448/3c46ddb1-5781-4367-bf31-b6f93678af7a Comparison.of.Short.Term.Partners.15-49.Male.Over.Time.png (view on web)https://github.com/UCL/hivpy/assets/34184448/faa1856e-f2e6-457b-8c51-b9b4e88d6373 Comparison.of.Short.Term.Partners.15-49.Female.Over.Time.png (view on web)https://github.com/UCL/hivpy/assets/34184448/a7a39165-1652-4664-bc9e-8621665e1363
Partner Balance
@mmcleod89https://github.com/mmcleod89 is currently in the process of making improvements to the partner balance regulation code so we don't need to take too much stock of these graphs, but having them here for later comparison won't hurt.
Comparison.of.Partner.Sex.Balance.15-24.Male.Over.Time.png (view on web)https://github.com/UCL/hivpy/assets/34184448/9cc67d2b-7904-41a3-90e8-3ea4c4e60369 Comparison.of.Partner.Sex.Balance.15-24.Female.Over.Time.png (view on web)https://github.com/UCL/hivpy/assets/34184448/17da86e9-ae46-4c76-b2b1-418c5b70cd28 Comparison.of.Partner.Sex.Balance.25-34.Male.Over.Time.png (view on web)https://github.com/UCL/hivpy/assets/34184448/9dd66384-d22b-40d1-aeee-ce5239ee9d35 Comparison.of.Partner.Sex.Balance.25-34.Female.Over.Time.png (view on web)https://github.com/UCL/hivpy/assets/34184448/28364db1-264b-40bb-869b-b94cf3f25043 Comparison.of.Partner.Sex.Balance.35-44.Male.Over.Time.png (view on web)https://github.com/UCL/hivpy/assets/34184448/0c426be2-4626-4e86-81db-a8ee58ec8505 Comparison.of.Partner.Sex.Balance.35-44.Female.Over.Time.png (view on web)https://github.com/UCL/hivpy/assets/34184448/d2178626-f3b3-4e00-a996-f2f48e20cf04
— Reply to this email directly, view it on GitHubhttps://github.com/UCL/hivpy/issues/175#issuecomment-1949292825, or unsubscribehttps://github.com/notifications/unsubscribe-auth/AJNUSMVAQA3RNWFQAZEMES3YT666RAVCNFSM6AAAAABCPMCLVGVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMYTSNBZGI4TEOBSGU. You are receiving this because you are on a team that was mentioned.Message ID: @.**@.>>
Hi @ValentinaCambiano, thanks for your comments.
Non-HIV deaths (ratio)
is trying to calculate the death rate. I decided to add this metric when I noticed how different the population sizes were between the HIVpy and SAS models to see if that accounted for the increased number of deaths in the SAS runs.Short term partners (15-49, male)
and Short term partners (15-49, female)
are imperfectly calculated from the SAS runs because there are no s_m_4549_newp
or s_w_4549_newp
columns, there are only s_m_4554_newp
or s_w_4554_newp
, so I tried to get as close as I could without their inclusion. The bigger problem seems to be that I forgot to exclude sex workers from the HIVpy Short term partners (15-49, female)
data, so I'll be sure to add that as a constraint in my next round of plots.
Direct comparisons between Python and SAS From 1989 to 1995