Learning to Call: A Field Trial of a Collaborative Bandit Algorithm for Improved Message Delivery in Mobile Maternal Health

2025-07-22

AI Summary

The Kilkari mobile health initiative in India delivers maternal health information via weekly voice calls, but random dialing results in low call completion rates and insufficient message delivery. Method: We propose a personalized call-time optimization framework based on collaborative multi-armed bandits, integrating Bayesian updating with cross-user knowledge transfer to jointly model individual answering preferences while respecting privacy and heterogeneity. Contribution/Results: This work represents the first large-scale deployment of a collaborative bandit algorithm for dynamic call scheduling in mobile health interventions. A field experiment involving 6,500 real users demonstrates statistically significant improvements in call completion rates. The approach exhibits strong scalability, robustness under resource constraints, and practical viability for low-infrastructure settings—validating both its efficacy and operational feasibility in real-world public health programs.

Technology Category

• Machine Learning: Online Learning & Bandits
• Search and Optimization: Sampling/Simulation-based Search
• Reasoning under Uncertainty: Stochastic Optimization

Application Category

• User Modeling, Personalization and Recommendation: On-Device user modeling, personalization, and recommendation
• Responsible Web: Human-perceived consequences of algorithmic deployment on the web
• Security and Privacy: Large-scale security measurements

Abstract

Mobile health (mHealth) programs utilize automated voice messages to deliver health information, particularly targeting underserved communities, demonstrating the effectiveness of using mobile technology to disseminate crucial health information to these populations, improving health outcomes through increased awareness and behavioral change. India's Kilkari program delivers vital maternal health information via weekly voice calls to millions of mothers. However, the current random call scheduling often results in missed calls and reduced message delivery. This study presents a field trial of a collaborative bandit algorithm designed to optimize call timing by learning individual mothers' preferred call times. We deployed the algorithm with around $6500$ Kilkari participants as a pilot study, comparing its performance to the baseline random calling approach. Our results demonstrate a statistically significant improvement in call pick-up rates with the bandit algorithm, indicating its potential to enhance message delivery and impact millions of mothers across India. This research highlights the efficacy of personalized scheduling in mobile health interventions and underscores the potential of machine learning to improve maternal health outreach at scale.

Problem

• Optimizing call timing to improve maternal health message delivery
• Reducing missed calls in mobile health programs using bandit algorithms
• Enhancing maternal health outreach through personalized call scheduling

Innovation

• Collaborative bandit algorithm optimizes call timing
• Personalized scheduling improves call pick-up rates
• Machine learning enhances maternal health outreach