Computers & Electrical Engineering ( IF 4.3 ) Pub Date : 2024-01-20 , DOI: 10.1016/j.compeleceng.2023.109071 Rongyu Dou , Xin Kang
This paper introduces TAM-SenticNet, a Neuro-Symbolic AI framework uniquely designed for early depression detection through social media content analysis. Merging neural networks for feature extraction and sentiment analysis with advanced symbolic reasoning, TAM-SenticNet overcomes the limitations of traditional diagnostic tools, particularly in real-time responsiveness and interpretability. The symbolic reasoning, powered by SenticNet, provides a deep, structured understanding of emotional expressions, greatly enhancing model explainability and logical inference. Empirical evaluations reveal that TAM-SenticNet excels beyond existing models in performance metrics, achieving a of 0.665, of 0.881, and -score of 0.758, coupled with superior latency metrics, including and at 0.025, at 1.0, and at 0.675. These achievements highlight TAM-SenticNet’s cutting-edge approach to early depression detection, making it a pioneering tool in the application of AI for mental health informatics.
中文翻译:
TAM-SenticNet:一种通过社交媒体分析进行早期抑郁症检测的神经符号人工智能方法
本文介绍了 TAM-SenticNet,这是一种神经符号人工智能框架,专为通过社交媒体内容分析检测早期抑郁症而设计。TAM-SenticNet 将用于特征提取和情感分析的神经网络与高级符号推理相结合,克服了传统诊断工具的局限性,特别是在实时响应性和可解释性方面。由 SenticNet 提供支持的符号推理提供了对情感表达的深入、结构化的理解,极大地增强了模型的可解释性和逻辑推理。实证评估表明,TAM-SenticNet 在性能指标方面优于现有模型,实现了0.665,0.881,以及- 得分为 0.758,加上出色的延迟指标,包括和为 0.025,为 1.0,并且为 0.675。这些成就凸显了 TAM-SenticNet 在早期抑郁症检测方面的前沿方法,使其成为人工智能在心理健康信息学应用中的先锋工具。