AI-Enhanced Text Messaging for Behavioral Health: Designing the Next Generation of Digital Interventions

Rachel Kornfield, PhD
Northwestern University

ABSTRACT:
Text messaging is a uniquely accessible way to deliver digital health interventions, offering high levels of reach and usability relative to most apps. When paired with cutting-edge machine learning techniques, text messaging can also be adaptive and highly personalized. In this talk, Dr. Kornfield will describe how her team is leveraging machine learning and artificial intelligence within Small Steps SMS, an automated text messaging program that supports self-management of depression and anxiety. She will highlight how the program uses reinforcement learning, a form of machine learning that enables systems to optimize decisions over time, allowing daily messages to adapt content based on user engagement patterns, mood, and availability. She will also discuss her ongoing work assessing how generative AI can be incorporated into a rule-based text messaging system to expand and customize content while maintaining safety and fidelity. Finally, she will discuss key lessons from implementing these tools in real-world settings, including how a partnership with Mental Health America has informed adaptations to Small Steps SMS. This includes co-developing content for underserved populations (e.g., adolescents, racial/ethnic minorities) and designing workflows that align with the operational needs of community-based organizations.