Conversational mHealth Platform Designed to Support Tuberculosis Treatment Adherence in Low-Income South African Patients: Pilot Cohort Study Examining Coverage, Engagement, and Treatment Outcomes (Preprint)
Bibliographic record
Abstract
Background: Tuberculosis is a leading cause of death in South Africa, with poor adherence undermining treatment success. Findings from recent research on the impact of mHealth (mobile health) interventions on tuberculosis treatment outcomes show promise, yet many interventions remain untested in African contexts. Rising smartphone ownership in South Africa enables more complex mHealth interventions, offering an opportunity to deploy behavioral tools within high-burden, resource-constrained settings. Objective: This pilot study evaluates the feasibility and effectiveness, among low-income patients at a South African clinic, of a WhatsApp (Meta)-based conversational mHealth platform designed to tackle specific behavioral barriers to adherence. Aims include the following: (1) evaluating coverage by studying the proportion of patients within the target group who own smartphones, (2) describing patterns of engagement with the platform and the role of mobile data scarcity as a barrier to use, and (3) producing evidence on the impact that a behavioral mHealth intervention can have on tuberculosis treatment success. Methods: Patients newly diagnosed with drug-susceptible pulmonary tuberculosis between August 2022 and October 2023 completed a screening survey. Those owning compatible mobile phones were invited to enroll. The platform provided reminders alongside behavioral support features. Coverage was studied by estimating smartphone ownership among screened patients and comparing characteristics between enrolled patients (n=42) and those receiving standard care (n=102) using standardized differences. Engagement was analyzed using local polynomial regressions for usage trends and logistic regressions to estimate the impact of mobile data top-ups. The marginal effect of enrollment on the probability of successfully completing tuberculosis treatment was studied using a t test and logistic regressions with and without covariates. Results: A total of 34% (49/146) of screened participants owned a phone that could use WhatsApp. There were differences in characteristics by enrollment status. Further, 50% of patients engaged with the platform each day until the end of treatment. Overcoming an initial inability to send unprompted messages to inactive patients was associated with an immediate 13-percentage-point increase in aggregate engagement the following month. Mobile data scarcity hindered use-receiving mobile data top-ups within the previous week was associated with a 3.37-percentage-point increase (95% CI 0.0007 to 0.0666) in platform engagement. The estimated marginal effect of enrollment was a 17.6-percentage-point (95% CI 0.003 to 0.348) increase in treatment completion, becoming attenuated after adjusting for patient characteristics (12.8 percentage points, 95% CI -0.048 to 0.304). Conclusions: While phone ownership and mobile data constraints represent barriers to feasibility, findings suggest that smartphone-based mHealth interventions may aid successful treatment completion-alleviating health system burdens by automating care for less vulnerable patients. Engagement with the platform throughout tuberculosis treatment was high and stable, and enrolled users experienced a higher success rate. A randomized controlled trial is required for impact evaluation.
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 | 0.004 | 0.013 |
| Meta-epidemiology (narrow) | 0.000 | 0.001 |
| Meta-epidemiology (broad) | 0.001 | 0.001 |
| Bibliometrics | 0.001 | 0.001 |
| Science and technology studies | 0.003 | 0.001 |
| Scholarly communication | 0.002 | 0.002 |
| Open science | 0.001 | 0.002 |
| Research integrity | 0.001 | 0.002 |
| Insufficient payload (model declined to judge) | 0.007 | 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".