A significant obstacle in developing trustworthy artificial intelligence systems for mental health research is the poor quality of annotations in depression-related datasets. These annotations often lack structured evidence, symptom-level justification, and alignment with the Diagnostic and Statistical Manual of Mental Disorders, Fifth Edition, Text Revision (DSM-5-TR) criteria1. To address this issue, researchers have proposed a self-evolving human-centered framework for explainable depression symptom annotation. This framework aims to improve annotation quality by providing a structured approach to labeling depression symptoms, ensuring that annotations are evidence-based and aligned with established diagnostic criteria. The framework's self-evolving nature allows it to adapt to new data and updates in diagnostic criteria, making it a potentially valuable tool for mental health research. This development matters to practitioners because it has the potential to enhance the reliability and explainability of AI systems used in mental health diagnosis and treatment.