MERID — Multimodal Recursive Self-Improvement Agents for Major Depression Analysis (arXiv:2609.36235v1)
This arXiv preprint introduces MERID, a framework that uses recursive self-improvement agents to iteratively develop multimodal pipelines for detecting and estimating severity of major depressive disorder. MERID combines Grounded State Construction (GSC), Coupled Pipeline Exploration (CPE), and Evidence-Guided Evolution (EGE), reports improved performance versus multimodal and agent-based baselines on depression benchmarks, and publishes code at GitHub: DiscoAILab/MERID.