A culturally-adapted AI conversational agent delivered sexual and reproductive health education on smartphones/tablets to 160 young women in rural Lesotho; a clinical trial found significant knowledge gains on 2 of 3 topics.
172 women aged 18-28
Women enrolled in the trial
160 women
Active Nthabi users
76.6 to 89.9 % correct
Family-planning knowledge, before vs after (up to 6 weeks)
45.3 to 71.6 % correct
Folic-acid knowledge, before vs after (up to 6 weeks)
69.5 to 72.6 % correct
Healthy-eating knowledge, before vs after (not significant) (up to 6 weeks)
91 %
Participants who were unmarried
Details
Maturity
Pilot
Promoter
Sefako Makgatho University of Health Sciences & Northeastern University (Gabby Lab)
Period
2021 to 2024
Keywords
digital health, conversational AI, sexual and reproductive health, mobile health
Context
Nthabi is a Sesotho-speaking, culturally adapted AI conversational health educator — modelled as a Mosotho nurse-midwife — delivered on smartphones/tablets to young women in rural Lesotho, adapting the US-developed 'Gabby' agent system through a five-institution research collaboration funded by the US NIH Fogarty International Center.
Objectives
Test whether a culturally localised conversational agent can improve sexual and reproductive health knowledge (family planning, folic acid, healthy eating) among young rural Lesotho women in a resource-constrained, low-connectivity setting.
Activities
A single-arm trial enrolled 172 women aged 18-28 in the rural districts of Leribe and Berea; 160 used Nthabi and 157 began interactions with the agent over up to six weeks of pre/post knowledge testing.
Results
Family-planning knowledge rose significantly from 76.6% to 89.9% correct (p=0.0233) and folic-acid knowledge rose from 45.3% to 71.6% correct (p<0.0001); healthy-eating knowledge rose only marginally, from 69.5% to 72.6%, and was not statistically significant (p=0.421).
Conclusions
Researchers reported the non-significant healthy-eating result honestly rather than omitting it; a companion study found the cultural adaptation was rated appropriate and acceptable, but the trial remains research-stage with no published evidence yet of routine deployment beyond the study population.
Implementation
Indicative cost
Low (< €50k)
Time to results
Medium (1–3 years)
Staffing & skills
Multi-institution research consortium: Sefako Makgatho University of Health Sciences, Northeastern University (Gabby Lab), Boston University/Boston Medical Center, UMass Chan Medical School, University of Oklahoma Health Sciences Center
Conditions for success
Cultural adaptation of the agent (Sesotho language, idioms, Mosotho nurse-midwife character appearance) rated appropriate and acceptable by participants in a companion study
Access to a smartphone or tablet for up to six weeks of interaction
Common failure modes
Healthy-eating knowledge gains were not statistically significant (69.5% to 72.6%, p=0.421), suggesting the conversational-agent format did not shift knowledge equally across all health topics
Do you run this practice?
Claim it —
verified implementers get a public contact pathway and can propose corrections.
Data sources
Where this practice's information was retrieved from, and when.
Pontificia Universidad Javeriana co-designed Youth Collective Minds, a digital mental-health platform for Bogotá's youth, with 146 participants; a 2022-23 field …
Blended digital-literacy training for 150 frontline immunization officers across 96 health centres in Sierra Leone raised computer-literacy proficiency from 9% …
