Preliminary Study of the Impact of AI-Based Interventions on Health and Behavioral Outcomes in Maternal Health Programs

TL;DR

This study addresses whether AI-driven scheduling of live interventions in the mMitra maternal health program can boost engagement with automated health messages and, in turn, improve postnatal health knowledge. It deploys a randomized design across three cohorts, employing a RMAB-based AI scheduler to select beneficiaries for live calls and then analyzes outcomes via a high-risk survey with matching-based comparisons. The results show a statistically significant increase in listenership (average +$7.43$ seconds per call over 12 calls; $p = 6.4656\times 10^{-5}$) and provide early evidence of improved behavioral/health knowledge among those who gained listenership, though overall health outcomes are not yet statistically conclusive due to sample size and noise. The findings support AI-driven resource allocation in mHealth and motivate larger-scale follow-up trials with refined cohorts and survey instruments to establish robust causal links to health outcomes.

Abstract

Automated voice calls are an effective method of delivering maternal and child health information to mothers in underserved communities. One method to fight dwindling listenership is through an intervention in which health workers make live service calls. Previous work has shown that we can use AI to identify beneficiaries whose listenership gets the greatest boost from an intervention. It has also been demonstrated that listening to the automated voice calls consistently leads to improved health outcomes for the beneficiaries of the program. These two observations combined suggest the positive effect of AI-based intervention scheduling on behavioral and health outcomes. This study analyzes the relationship between the two. Specifically, we are interested in mothers' health knowledge in the post-natal period, measured through survey questions. We present evidence that improved listenership through AI-scheduled interventions leads to a better understanding of key health issues during pregnancy and infancy. This improved understanding has the potential to benefit the health outcomes of mothers and their babies.

Figures (7)

• Figure 1: Questions asked in the survey.
• Figure 2: Score on single choice questions for beneficiaries who gained quartiles in listenership post-interventions and their corresponding beneficiaries from the control arm. The subset of beneficiaries from the intervention arm in all questions. The error bars represent the standard error in the measurement.
• Figure 3: Score on multiple choice questions for beneficiaries who gained quartiles in listenership post-interventions and their corresponding beneficiaries from the control arm. Except on Question $5$, beneficiaries from the intervention arm perform on average better.
• Figure 4: Score on groupings of questions for beneficiaries who gained quartiles in listenership and their corresponding beneficiaries from the control arm. Average scores in all categories are higher for beneficiaries from the intervention arm.
• Figure 5: Average score on single choice questions for all beneficiaries.