HIV prevalence — prior vs posterior, by country
Same Beta-Binomial machinery, four different priors. Lesotho's posterior reflects 4,700+ tagged CHW interactions; the others are prior-only — that's where deployment goes next.
Why these four
All four have stipended CHW cadres, WhatsApp penetration, and a TB+HIV+MNCH burden that justifies the same protocol corpus. Sources cited per row.
Why each country comes next
Lesotho (live). Pioneered SSA CHW institutionalisation — 11,000 Village Health Workers professionalised under the 2014 health reform. Highest TB-HIV co-infection rate. Government motivated, geography rural — perfect beachhead.
Botswana (next, ≤6 months). Wealthier health system, faster procurement. 95% ART coverage means treatment-cascade indicators dominate the conversation — precisely what luma's structured extraction surfaces. Setswana ASR shares phonetic structure with Sesotho — the voice pipeline retunes in days.
Eswatini (next, ≤9 months). Highest HIV prevalence on Earth (27%+). Maternal mortality crisis. Tiny, concentrated population (1.2M) means rapid coverage. Government enrolled in pan-African 2-million-CHW initiative — political will is already there. siSwati is mutually intelligible with isiZulu and Sesotho.
Malawi (moonshot, ≤18 months). 21.7M people, <0.5 doctors per 1,000. The 11,000 Health Surveillance Assistants operate under severe supervision deficit — luma is a force-multiplier. Chichewa is a high-resource ASR target. The largest single growth lever in our 36-month plan.
Same framework, swap the prior
The Beta-Binomial conjugate update doesn't care which country it's in. The prior changes (UNAIDS 18.5% in Lesotho, 19.7% in Botswana, 27% in Eswatini, 7.1% in Malawi). The CHW-observed counts change. The posterior reflects whichever combination is in front of it. Read the methodology →