"What is the best method of family planning for me?": a text mining analysis of messages between users and agents of a digital health service in Kenya
Bibliographic record
Abstract
<ns5:p> <ns5:bold>Background</ns5:bold> : Text message-based interventions have been shown to have consistently positive effects on health improvement and behavior change. Some studies suggest that personalization, tailoring, and interactivity can increase efficacy. With the rise in artificial intelligence and its incorporation into interventions, there is an opportunity to rethink how these characteristics are designed for greater effect. A key step in this process is to better understand how users engage with interventions. In this paper, we apply a text mining approach to characterize the ways that Kenyan men and women communicated with the first iterations of <ns5:italic>askNivi</ns5:italic> , a free sexual and reproductive health information service. </ns5:p> <ns5:p> <ns5:bold>Methods</ns5:bold> : We tokenized and processed more than 179,000 anonymized messages that users exchanged with live agents, enabling us to count word frequency overall, by sex, and by age/sex cohorts. We also conducted two manual coding exercises: (1) We manually classified the intent of 3,834 user messages in a training dataset; and (2) We manually coded all conversations between a random subset of 100 users who engaged in extended chats. </ns5:p> <ns5:p> <ns5:bold>Results</ns5:bold> : Between September 2017 and January 2019, 28,021 users (mean age 22.5 years, 63% female) sent 87,180 messages to <ns5:italic>askNivi,</ns5:italic> and 18 agents sent 92,429 replies. Users wrote most often about family planning methods, contraception, side effects, pregnancy, menstruation, and sex, but we observed different patterns by sex and age. User intents largely reflected the marketing focus on reproductive health, but other topics emerged. Most users sought factual information, but requests for advice and symptom reports were common. </ns5:p> <ns5:p> <ns5:bold>Conclusions</ns5:bold> : Young people in Kenya have a great desire for accurate and reliable information on health and wellbeing, which is easy to access and trustworthy. Text mining is one way to better understand how users engage with interventions like <ns5:italic>askNivi</ns5:italic> and maximize what artificial intelligence has to offer. </ns5:p>
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.001 | 0.007 |
| Meta-epidemiology (narrow) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.000 |
| Bibliometrics | 0.002 | 0.003 |
| Science and technology studies | 0.002 | 0.001 |
| Scholarly communication | 0.001 | 0.002 |
| Open science | 0.001 | 0.001 |
| Research integrity | 0.001 | 0.001 |
| Insufficient payload (model declined to judge) | 0.001 | 0.001 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".