- 學位論文
利用深度學習檢測憂鬱症與情緒輔助之社交機器人
Deep Learning based Depression Detection and Emotional Support with Social Assistant Robot
本文將於2026/08/01開放下載。若您希望在開放下載時收到通知，可將文章加入收藏。
摘要
憂鬱症是一種常見的心理疾病，對患者的日常生活產生了很大的影響。然而， 由於社會文化或經濟壓力等原因，許多人可能不敢尋求心理諮詢，從而導致憂鬱 症問題加劇或延誤就診時間。因此，開發一個能夠隨時檢測自身心理狀態的憂鬱 症偵測系統至關重要。本論文提出了一種基於深度學習技術實現的憂鬱症偵測模 型，利用社交型機器人與使用者進行交流，提供情緒支持和安撫的回應方式，讓 使用者能更自由地表達內心感受。 目前，大多數與憂鬱症相關的方法主要關注於憂鬱症偵測的預測能力，但未 能考慮實際應用場景。因此，本論文提出了一種基於圖神經網路的架構，該架構 利用使用者完整的對話紀錄作為輸入，結合單詞和句子的多方面語義特徵，同時 整合了情緒和生理特徵等額外資訊，以更全面地預測憂鬱症。此外，在對話過程 中，系統以具有情緒輔助能力的回覆方式與使用者進行交流。當對話結束時，系 統根據對話內容生成對應的偵測結果，並以自然語言的方式回覆，以提高使用者 與系統之間的流暢性。因此，本論文提出的社交型機器人不僅能檢測憂鬱症，還 能提供情緒安撫和支持，使使用者更好地了解自己的心理狀態。
並列摘要
Depression is a common mental disorder that has a significant impact on the day-to-day life of individuals. Due to social, cultural, or economic pressures, many people are reluctant to seek psychological counseling, resulting in worsening depression or delayed treatment. Therefore, it is crucial to develop a depression detection system. This system should be able to monitor one’s psychological state at any time. In this thesis, a depression detection model based on deep learning techniques is proposed, which uses a social robot to engage in conversations with users and to provide emotionally supportive and comforting responses, thus allowing users to freely express their inner feelings. Currently, most of the depression detection methods mainly focus on the predictive capability of depression detection without consideration of practical application scenarios. Therefore, an architecture based on graph neural networks using complete user conversation records as input is presented in this thesis To comprehensively predict depression, this architecture integrates various semantic features of words and sentences as well as additional information such as emotional and biological characteristics. In addition, the system interacts with the user during the call by responding with emotionally comforting responses. At the end of the conversation, the system generates appropriate recognition results based on the content of the conversation and replies in natural language to enhance the communication between the user and the system. In this way, the social assistant robot proposed in this thesis not only detects depression but also provides emotional support and comfort, enabling the user to have a better understanding of his or her mental state.
參考文獻
Siyang Liu, Chujie Zheng, Orianna Demasi, Sahand Sabour, Yu Li, Zhou Yu, Yong Jiang, and Minlie Huang. Towards emotional support dialog systems. In Proceedings of the 59th Annual Meeting of the Association for Computational Linguistics and the 11th International Joint Conference on Natural Language Processing (Volume 1: Long Papers), pages 3469–3483, 2021.
World Health Organization. Depressive disorder (depression). https://www.who.int/news-room/fact-sheets/detail/depression (accessed May 25, 2022).
Shoji Yokoya, Takami Maeno,Naoto Sakamoto,Ryohei Goto,and Tetsuhiro Maeno. A brief survey of public knowledge and stigma towards depression. Journal of clinical medicine research, 10(3):202, 2018.
Anja Thieme, Danielle Belgrave, and Gavin Doherty. Machine learning in mental health: A systematic review of the hci literature to support the development of effective and implementable ml systems. ACM Transactions on Computer-Human Interaction (TOCHI), 27(5):1–53, 2020.
Chuyuan Li, Chloé Braud, and Maxime Amblard. Multi-task learning for depression detection in dialogs. In Proceedings of the 23rd Annual Meeting of the Special Interest Group on Discourse and Dialogue, pages 68–75, 2022.