Self-Evolving Human-Centered Framework for Explainable Depression Symptom Annotation

2026-07-16

AI Summary

Current depression datasets lack structured symptom-level annotations aligned traceably with DSM-5-TR criteria, limiting the interpretability and reliability of AI systems. This work proposes a human-AI collaborative, self-evolving annotation framework that integrates large language models with expert-in-the-loop oversight through a three-stage structured pipeline to generate DSM-5-TR-compliant symptom annotations. The framework incorporates a dual-memory architecture—comprising example memory and reflection memory—to internalize expert feedback, enabling continuous annotation refinement without retraining. Pilot studies demonstrate that this approach significantly improves annotation consistency and interpretability, reduces manual revision costs, and produces auditable clinical evidence alongside complete reasoning traces.

Abstract

Annotation quality is a major bottleneck in building reliable and explainable artificial intelligence (XAI) systems for mental health research. In depression-related datasets, labels are often assigned without structured evidence, symptom-level justification, or traceable alignment with the criteria of the Diagnostic and Statistical Manual of Mental Disorders, Fifth Edition, Text Revision (DSM-5-TR), limiting both transparency and downstream model interpretability. We propose a self-evolving, expert-in-the-loop annotation framework for Major Depressive Disorder (MDD) that combines large language model (LLM)-assisted labeling with expert verification. The framework is intended to support the construction of explainable, DSM-5-TR-aligned datasets rather than to perform clinical diagnosis. It operates in three stages: candidate evidence selection from textual records, criterion-level DSM-5-TR analysis, and case-level synthesis that produces label-level diagnostic and severity annotations. A dual-memory architecture, composed of Example Memory and Reflection Memory, is designed to internalize expert feedback and iteratively improve future annotations without retraining. We describe this mechanism and leave its evaluation across multiple feedback cycles to future work. In addition to final labels, the framework exports clinical evidence, reasoning traces, and edit histories, enabling comprehensive auditability. In a pilot study using expert-reviewed samples, the proposed approach improves annotation consistency and explainability while reducing manual revision effort.

Problem

• annotation quality
• explainable AI
• depression symptom annotation
• DSM-5-TR alignment
• mental health datasets

Innovation

• self-evolving framework
• explainable AI
• DSM-5-TR alignment
• dual-memory architecture
• expert-in-the-loop annotation