JMIR Publications

ABSTRACT

Background:

India faces a dual burden of diabetes and prediabetes, and mobile health (mHealth) interventions have shown promise in promoting healthy lifestyle changes.

Objective:

We aimed to evaluate the effectiveness of an AI-enabled mHealth intervention compared to traditional mHealth in fostering diabetes prevention behaviors in the rural district of Gulbarga, Karnataka, South India.

Methods:

A quasi-experimental pre-post study was conducted among adult without diabetes (n=1048) in Gulbarga. Participants were divided into intervention and control groups. The control group received static diabetes prevention messages via WhatsApp, while the intervention group received customized messages twice a week based on individual feedback through reinforcement learning algorithms. Data was collected via home interviews on demographics, diabetes knowledge, and lifestyle behaviors. Chi-square tests and t-tests were used to assess group differences. Intervention effects were evaluated using multivariable logistic regression for binary outcomes and ANCOVA for continuous outcomes. Adjusted odds ratios with 95% confidence intervals were reported, and Bonferroni correction was applied for multiple comparisons.

Results:

Of the 1,082 participants enrolled, 1,048 (97%) completed the 6 month follow-up, more than one third were female (n=661, 63.1%). At endline, no significant differences were observed between groups for primary outcomes. Physical activity (≥30 minutes/day) at endline, the aOR was 1.0 (95% CI 0.7–1.3, p = 0.74), and baseline activity (aOR = 2.1, 95% CI 1.5–3.1, p < 0.001) and age >50 years (aOR = 3.8, 95% CI 1.6–9.3, p = 0.003) being significant predictors for endline physical activity. Being employed was associated with lower odds of physical activity (aOR = 0.2, 95% CI 0.1–0.3, p < 0.001). Daily fruit intake, the aOR was 1.4 (95% CI 0.8–2.3, p = 0.24). Participants aged 26–35 years had higher odds of daily fruit intake (aOR = 4.7, 95% CI 1.9–11.8, p = 0.001), while employment was associated with lower odds (aOR = 0.3, 95% CI 0.1–0.8, p = 0.022). Mean BMI difference at endline was –0.0 kg/m² (95% CI –0.6 to 0.5, p = 0.95). Baseline BMI was a strong predictor of endline BMI (p < 0.001). Exploratory behavioral outcomes revealed no significant differences: stair use (aOR = 0.9, 95% CI 0.7–1.4, p = 0.79), walking for chores (aOR = 2.4, 95% CI 1.0–6.1, p = 0.06), helping with household chores (aOR = 1.0, 95% CI 0.4–2.3, p = 0.94), and farm work (aOR = 1.3, 95% CI 0.9–1.8, p = 0.19).

Conclusions:

Both AI-enabled and traditional mHealth interventions demonstrated similar effectiveness in promoting diabetes prevention behaviors in rural India. The lack of significant differences between approaches suggests that well-designed traditional mHealth messaging may be as effective as AI-customized interventions for diabetes prevention.