Few-Shot Learning for Mental Disorder Detection: A Continuous Multi-Prompt Engineering Approach with Medical Knowledge Injection

2024-01-16

AI Summary

To address high annotation costs and poor model generalizability in mental disorder text screening, this paper proposes a continuous multi-prompt engineering framework based on large language models (LLMs). The method jointly models user-specific behavioral patterns and structured medical knowledge—such as knowledge graph embeddings—via a novel continuous multi-prompt optimization paradigm, enabling dynamic, task-agnostic integration of personalized features and domain knowledge without relying on task-specific architectures or discrete prompts. Evaluated on three common mental disorder detection tasks, the approach achieves high accuracy using only 2–5 positive samples per class, substantially outperforming conventional feature engineering, architecture engineering, and discrete prompting methods. Moreover, when transferred across disorders to rare conditions, it improves AUC by over 12%, significantly enhancing few-shot generalization capability and clinical applicability.

Technology Category

• Machine Learning: Large Multimodal Models (LMMs)
• Natural Language Processing: Prompt Engineering / Prompting
• Search and Optimization: Mixed Discrete/Continuous Search

Application Category

• User Modeling, Personalization and Recommendation: Large Language Models (LLM) for user modeling and recommendation
• Economics, Online Markets and Human Computation: Humans versus LLMs for data annotation and labeling
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs

Abstract

This study harnesses state-of-the-art AI technology for detecting mental disorders through user-generated textual content. Existing studies typically rely on fully supervised machine learning, which presents challenges such as the labor-intensive manual process of annotating extensive training data for each research problem and the need to design specialized deep learning architectures for each task. We propose a novel method to address these challenges by leveraging large language models and continuous multi-prompt engineering, which offers two key advantages: (1) developing personalized prompts that capture each user's unique characteristics and (2) integrating structured medical knowledge into prompts to provide context for disease detection and facilitate predictive modeling. We evaluate our method using three widely prevalent mental disorders as research cases. Our method significantly outperforms existing methods, including feature engineering, architecture engineering, and discrete prompt engineering. Meanwhile, our approach demonstrates success in few-shot learning, i.e., requiring only a minimal number of training examples. Moreover, our method can be generalized to other rare mental disorder detection tasks with few positive labels. In addition to its technical contributions, our method has the potential to enhance the well-being of individuals with mental disorders and offer a cost-effective, accessible alternative for stakeholders beyond traditional mental disorder screening methods.

Problem

• Detects mental disorders using user-generated text with AI.
• Reduces need for extensive labeled data and specialized architectures.
• Enables few-shot learning for rare mental disorder detection.

Innovation

• Utilizes large language models for mental disorder detection.
• Implements continuous multi-prompt engineering with medical knowledge.
• Achieves few-shot learning with minimal training examples.