PanicToCalm: A Proactive Counseling Agent for Panic Attacks

TL;DR

This paper introduces PACE, a novel dataset for panic attack intervention, and PACER, a counseling model trained on this data, demonstrating improved effectiveness and human preference over existing models in crisis scenarios.

Contribution

The paper presents PACE, a first-of-its-kind dataset for panic attack counseling, and PACER, a specialized model optimized for empathetic crisis intervention.

Findings

PACER outperforms baseline models in counseling quality metrics.

Human evaluations favor PACER over other models in panic scenarios.

PACER shows significant improvement in client affect and crisis management.

Abstract

Panic attacks are acute episodes of fear and distress, in which timely, appropriate intervention can significantly help individuals regain stability. However, suitable datasets for training such models remain scarce due to ethical and logistical issues. To address this, we introduce PACE, which is a dataset that includes high-distress episodes constructed from first-person narratives, and structured around the principles of Psychological First Aid (PFA). Using this data, we train PACER, a counseling model designed to provide both empathetic and directive support, which is optimized through supervised learning and simulated preference alignment. To assess its effectiveness, we propose PanicEval, a multi-dimensional framework covering general counseling quality and crisis-specific strategies. Experimental results show that PACER outperforms strong baselines in both counselor-side metrics and client affect improvement. Human evaluations further confirm its practical value, with PACER consistently preferred over general, CBT-based, and GPT-4-powered models in panic scenarios (Code is available at https://github.com/JihyunLee1/PanicToCalm ).