Increasing engagement with cognitive-behavioral therapy (CBT) using generative AI: a randomized controlled trial...

Area of Science:

• Digital mental health
• Artificial intelligence in healthcare
• Cognitive behavioral therapy

Background:

• Mental healthcare shortages create significant unmet patient needs.
• Digital interventions offer scalable solutions but struggle with patient engagement.
• Generative AI (genAI) presents an opportunity to enhance engagement and personalization in digital mental health treatments.

Purpose of the Study:

• To evaluate if a genAI-enabled cognitive behavioral therapy (CBT) app improves engagement and reduces symptoms compared to digital CBT workbooks.
• To assess the safety and efficacy of genAI in a self-guided digital mental health intervention.

Main Methods:

• A randomized controlled trial (N=540) comparing a genAI app to digital CBT workbooks over six weeks.
• Participants with elevated anxiety or depression symptoms were recruited online.
• Primary outcomes included engagement frequency/duration and changes in GAD-7 and PHQ-9 scores.

Main Results:

• The genAI app increased engagement frequency (2.4x) and duration (3.8x) compared to workbooks.
• No significant differences in anxiety or depression symptom reduction were observed between groups.
• Exploratory analysis indicated stronger anxiety reduction with genAI personalization features.

Conclusions:

• GenAI-enabled therapy safely enhances user engagement in self-directed use compared to static digital materials.
• Current findings do not show overall enhancement in symptom reduction for anxiety or depression.
• GenAI holds potential for scalable, engaging digital mental health interventions.