Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Human Empathy as Encoder: AI-Assisted Depression Assessment in Special Education

Boning Zhao, Xinnuo Li, Yutong Hu

TL;DR

The paper tackles depression assessment in sensitive special-education environments, where standardized questionnaires and purely automated analyses can miss nuanced states. It proposes Human Empathy as Encoder (HEAE), a framework that structurally encodes teachers’ empathetic insights as a 9-dimensional Empathy Vector (EV) guided by PHQ-9, and fuses it with student narratives via Asymmetric Cross-Modal Enhancement (ACME). The results show a final model performance of 82.05% accuracy (±0.58) and 82.08% macro F1 for 7-level depression severity, demonstrating improved accuracy over baselines while preserving transparency and privacy through local processing. The work contributes a practical blueprint for socially responsible, human-centered affective computing in education, with implications for explainability, ethics, and real-world deployment.

Abstract

Assessing student depression in sensitive environments like special education is challenging. Standardized questionnaires may not fully reflect students' true situations. Furthermore, automated methods often falter with rich student narratives, lacking the crucial, individualized insights stemming from teachers' empathetic connections with students. Existing methods often fail to address this ambiguity or effectively integrate educator understanding. To address these limitations by fostering a synergistic human-AI collaboration, this paper introduces Human Empathy as Encoder (HEAE), a novel, human-centered AI framework for transparent and socially responsible depression severity assessment. Our approach uniquely integrates student narrative text with a teacher-derived, 9-dimensional "Empathy Vector" (EV), its dimensions guided by the PHQ-9 framework,to explicitly translate tacit empathetic insight into a structured AI input enhancing rather than replacing human judgment. Rigorous experiments optimized the multimodal fusion, text representation, and classification architecture, achieving 82.74% accuracy for 7-level severity classification. This work demonstrates a path toward more responsible and ethical affective computing by structurally embedding human empathy

Human Empathy as Encoder: AI-Assisted Depression Assessment in Special Education

TL;DR

The paper tackles depression assessment in sensitive special-education environments, where standardized questionnaires and purely automated analyses can miss nuanced states. It proposes Human Empathy as Encoder (HEAE), a framework that structurally encodes teachers’ empathetic insights as a 9-dimensional Empathy Vector (EV) guided by PHQ-9, and fuses it with student narratives via Asymmetric Cross-Modal Enhancement (ACME). The results show a final model performance of 82.05% accuracy (±0.58) and 82.08% macro F1 for 7-level depression severity, demonstrating improved accuracy over baselines while preserving transparency and privacy through local processing. The work contributes a practical blueprint for socially responsible, human-centered affective computing in education, with implications for explainability, ethics, and real-world deployment.

Abstract

Assessing student depression in sensitive environments like special education is challenging. Standardized questionnaires may not fully reflect students' true situations. Furthermore, automated methods often falter with rich student narratives, lacking the crucial, individualized insights stemming from teachers' empathetic connections with students. Existing methods often fail to address this ambiguity or effectively integrate educator understanding. To address these limitations by fostering a synergistic human-AI collaboration, this paper introduces Human Empathy as Encoder (HEAE), a novel, human-centered AI framework for transparent and socially responsible depression severity assessment. Our approach uniquely integrates student narrative text with a teacher-derived, 9-dimensional "Empathy Vector" (EV), its dimensions guided by the PHQ-9 framework,to explicitly translate tacit empathetic insight into a structured AI input enhancing rather than replacing human judgment. Rigorous experiments optimized the multimodal fusion, text representation, and classification architecture, achieving 82.74% accuracy for 7-level severity classification. This work demonstrates a path toward more responsible and ethical affective computing by structurally embedding human empathy

Paper Structure

This paper contains 34 sections, 1 equation, 5 figures, 1 table.

Table of Contents

  1. Introduction
  2. Related Work
  3. Natural Language Processing for Depression Detection
  4. Multimodal Approaches in Mental Health Assessment
  5. Human-in-the-Loop (HITL) and Human-AI Collaboration
  6. Explainability, Ethics, and Responsibility in Affective Computing
  7. Positioning the Current Work
  8. Methodology
  9. Conceptual Framework: The Human Empathy as Encoder
  10. Dataset Generation and Features
  11. Golden Seed Development
  12. Novel Automated Annotation Pipeline
  13. Dataset Finalization and Characteristics
  14. Model Inputs, Target
  15. Final Model Architecture
  16. ...and 19 more sections

Figures (5)

  • Figure 1: Automated annotation pipeline using golden seeds and an LLM.
  • Figure 2: The architecture of the proposed Human Empathy as Encoder (HEAE) model.
  • Figure 3: Performance comparison of different multimodal fusion strategies.
  • Figure 4: Comparison of different pooling strategies for aggregating text chunk embeddings.
  • Figure 5: Performance comparison of different classifier head architectures.