Identifying Effective Strategies in Prompt Evolution for Stress and Emotion Prediction

Key points

• Early stress detection and risk prediction are crucial for timely mental health assessment and support.1
• While large language models have recently shown promise for social media stress prediction, relying on accuracy can obscure false positives.1
• In a separate set of searches, binary stress prediction improved early, whereas multilabel emotion prediction initially declined and partly recovered.1
• These findings identify effective prompt evolution methods, strategy combinations associated with gains, and task and evaluation limits, providing guidance for applying prompt evolution methods to social media stress detection.1

Sources (1)

• [1]Identifying Effective Strategies in Prompt Evolution for Stress and Emotion Prediction
  arXiv (AI, ML, NLP, CV, robotics, multi-agent) · Oct 15, 07:24 AM

  By comparing six methods on social media stress and emotion data across two model sizes, we found that GEPA and its extension combining instructions from two prompts (GEPA+merge) improved prediction over initial prompts in several settings.

  Early stress detection and risk prediction are crucial for timely mental health assessment and support.

Extractive summary: sentences quoted from the sources.