Overview
Heart disease remains the leading cause of death for women in the United States, underscoring the need for scalable, effective education strategies. Recent quasi-experimental analyses compare two delivery formats for heart attack education: a human-delivered SMS intervention and an automated AI chatbot named HeartBot. The aim is to determine how each format influences knowledge, recognition of symptoms, and emergency response actions among community-dwelling women without a prior heart disease history.
Two Intervention Pathways
The study leveraged two datasets collected between 2022 and 2024. Phase 1 involved a human interventionist delivering educational content via two online conversations over two days. Phase 2 introduced HeartBot, a 24/7 automated SMS education tool built to deliver the same core messages with a simplified, scripted conversation. Both datasets relied on similar eligibility criteria: women aged 25 or older, U.S. residents with internet and SMS access, no history of heart disease or cognitive impairment, and not health professionals or students.
Primary Outcomes: Knowledge and Response to a Heart Attack
Knowledge and awareness were assessed using four questions on a 1–4 scale, focusing on recognizing signs and symptoms, distinguishing heart attack symptoms from other issues, calling emergency services, and reaching an emergency department within 60 minutes. Across both delivery formats, participants showed significant knowledge gains after the interactions, indicating that structured educational dialogue—whether human-led or AI-guided—can improve critical heart attack literacy in women.
Effect Sizes and Context
In the human-delivered group (Phase 1), the adjusted odds of improved knowledge were notably high across all questions (e.g., recognizing signs and symptoms: AOR ≈ 15). The HeartBot group (Phase 2) also demonstrated meaningful improvements, though the effect sizes were smaller (e.g., recognizing signs and symptoms: AOR ≈ 7). A direct test comparing the two formats suggested humans provided stronger gains for most outcomes, though HeartBot remained significantly effective, particularly for recognizing symptoms and distinguishing heart attack signs.
User Experience and Conversation Quality
Beyond knowledge, researchers examined message effectiveness, humanness, naturalness, and coherence using established scales. Participants in the human-delivered study generally rated conversations more positively across these dimensions. Yet, a substantial portion of participants could not reliably tell whether they were interacting with a human or AI, highlighting evolving perceptions of machine-turned-partner in health education. The AI bot performed well on factual delivery, but many users perceived the human assistant as more natural and engaging.
<h2Interpretation: What Do These Findings Mean for Public Health?
The results point to several key takeaways. First, both human and AI chatbot formats can effectively raise knowledge about heart attack symptoms and appropriate responses, offering scalable options for broad outreach. Second, human-delivered conversations may yield stronger short-term gains and richer relational experiences, likely due to longer engagement sessions and more adaptive dialogue. Third, HeartBot demonstrates promising potential as a cost-effective, 24/7 educational tool that can reach large populations when human resources are limited.
Design Implications for Future Education Programs
To optimize heart attack education for women, programs should consider a hybrid approach: sustained, multi-session human-led sessions to deepen understanding, complemented by AI chatbot episodes to reinforce learning and extend reach. Improvements to HeartBot could include longer, adaptive conversations, more topics, and integrated quizzes to bolster retention. Importantly, robust randomized controlled trials are needed to establish causality and determine long-term effects on knowledge retention and real-world emergency behavior.
Limitations and Next Steps
Limitations include nonrandomized design, unequal session lengths, and different topic coverage between the two phases. Word-count disparities prevented definitive adjustment for dose effects. Recruitment through social media may bias toward digitally literate populations. Future research should address these gaps with randomized trials, diverse recruitment strategies, and longer follow-up to assess sustained knowledge and behavior changes.
Conclusion
The comparative analysis of human-delivered and AI chatbot education suggests both modalities can enhance women’s knowledge about heart attack recognition and response in the United States. HeartBot’s scalability and 24/7 availability make it a compelling option for public health campaigns, especially where human resources are constrained. However, to maximize impact, educators should explore hybrid designs that combine deep, interactive human dialogue with automated reinforcement to achieve durable learning outcomes.
