大多數人對憂鬱症的了解不深，認為只是心情不好，不需要治療，只要順其自然便會改善。在長期的壓力累積下，人們對於負面情緒的察覺力越來越低。若能有機制適度提醒使用者進行紓壓、平復情緒，有助於避免憂鬱情緒長時間累積。
本研究與精神科醫師合作建立一個情緒偵測及回饋系統的APP，利用手機螢幕顯示的應用程式名稱作為手機使用行為的基本資料。當系統辨識到負面情緒或預判出感受降低時，主動反饋醫師所提供的建議及可能造成的因子，可以讓使用者察覺自身狀況並適時改善情緒。我們利用自定義的Bayesian network架構為每位使用者建立個人化模組，用來預測使用者的情緒並依照CPTs計算最有可能影響使用者的因子作為回饋給使用者。透過14天的訓練資料收集後，系統可以透過過去30分鐘內的手機行為特徵及環境因子來做情緒的偵測。我們同時蒐集了輕度憂鬱症患者及一般使用者的資料。實驗的平均準確度為54%，從中也能發現病患平均使用時間高於一般使用者，且在各種時間間隔下的準確度都優於一般使用者。與分類器的比較下，最佳準確度僅僅輸給Naïve Bayes不到1%，Bayesian network非黑箱的優勢則遙遙領先分類器。

Most people do not quite understand about the depression, and they think that is just in a bad mood and do not need any treatment. In the long-term accumulation of pressure, the people are harder and harder to perceive their own negative emotion. If there is an appropriate mechanism to remind the user to do some activities, relieve pressure, and calm emotions, it's helpful for avoiding long melancholy mood accumulation.
This study cooperates with professional psychiatrists to develop a detection system on APPs, and using the package name of APP which displayed on screen as the basic information about smartphone using behaviors. When the system recognizes the negative mood, it would give appropriate feedback and help the user to ease their negative emotions and realize the reason that may cause the negative emotions. We use Bayesian network structure, which we defined, for each user to create personal model, and then use it to predict the user's mood and calculate the most likely factor as a feedback. After a 14-day personalized training period, it is able to detect emotional states based on the smartphone usage patterns at last 30 minutes. The average accuracy of the experiment is 54%, and we can also find that the patients' average accuracy and average using time on smartphone both are higher than normal users in any time slot. Comparing with the case of the classifiers, the best accuracy is only less than Naïve Bayes 1%, and the advantage of Bayesian network is not a black-box classification.

Beranuy, M., Oberst, U., Carbonell, X., & Chamarro, A. (2009). Problematic Internet and mobile phone use and clinical symptoms in college students: The role of emotional intelligence. Computers in human behavior, 25(5), 1182-1187.
Csikszentmihalyi, M., & Larson, R. (2014). Validity and reliability of the experience-sampling method Flow and the foundations of positive psychology (pp. 35-54): Springer.
Gil, G. B., Berlanga, A., & Molina, J. M. (2012). EmotionContext: user emotion dataset using smartphones. Paper presented at the International Workshop on Ambient Assisted Living.
Grossman, D., & Domingos, P. (2004). Learning Bayesian network classifiers by maximizing conditional likelihood. Paper presented at the Proceedings of the twenty-first international conference on Machine learning.
Hayamizu, T., Mutsuo, S., Miyawaki, K., Mori, H., Nishiguchi, S., & Yamashita, N. (2012). Group emotion estimation using Bayesian network based on facial expression and prosodic information. Paper presented at the Control System, Computing and Engineering (ICCSCE), 2012 IEEE International Conference on.
Kross, E., Verduyn, P., Demiralp, E., Park, J., Lee, D. S., Lin, N., . . . Ybarra, O. (2013). Facebook use predicts declines in subjective well-being in young adults. PloS one, 8(8), e69841.
Lamine, F. B., Kalti, K., & Mahjoub, M. A. (2012). The threshold EM algorithm for parameter learning in bayesian network with incomplete data. arXiv preprint arXiv:1204.1681.
Lane, R. D., Chua, P. M., & Dolan, R. J. (1999). Common effects of emotional valence, arousal and attention on neural activation during visual processing of pictures. Neuropsychologia, 37(9), 989-997.
LiKamWa, R., Liu, Y., Lane, N. D., & Zhong, L. (2013). MoodScope: building a mood sensor from smartphone usage patterns. Paper presented at the Proceeding of the 11th annual international conference on Mobile systems, applications, and services.
Mehrabian, A. (1996). Pleasure-arousal-dominance: A general framework for describing and measuring individual differences in temperament. Current Psychology, 14(4), 261-292.
Morris, M. E., Kathawala, Q., Leen, T. K., Gorenstein, E. E., Guilak, F., DeLeeuw, W., & Labhard, M. (2010). Mobile therapy: case study evaluations of a cell phone application for emotional self-awareness. Journal of medical Internet research, 12(2), e10.
Reis, H. T., & Gable, S. L. (2000). Event-sampling and other methods for studying everyday experience. Handbook of research methods in social and personality psychology, 190-222.
Ren, F., & Kang, X. (2013). Employing hierarchical Bayesian networks in simple and complex emotion topic analysis. Computer Speech & Language, 27(4), 943-968.
Sano, A., & Picard, R. W. (2013). Stress recognition using wearable sensors and mobile phones. Paper presented at the Affective Computing and Intelligent Interaction (ACII), 2013 Humaine Association Conference on.
Siegert, I., Böck, R., Vlasenko, B., Philippou-Hübner, D., & Wendemuth, A. (2011). Appropriate emotional labelling of non-acted speech using basic emotions, geneva emotion wheel and self assessment manikins. Paper presented at the 2011 IEEE International Conference on Multimedia and Expo.
Valenza, G., Nardelli, M., Lanata, A., Gentili, C., Bertschy, G., Paradiso, R., & Scilingo, E. P. (2014). Wearable monitoring for mood recognition in bipolar disorder based on history-dependent long-term heart rate variability analysis. IEEE Journal of Biomedical and Health Informatics, 18(5), 1625-1635.
Whittaker, R., Merry, S., Stasiak, K., McDowell, H., Doherty, I., Shepherd, M., . . . Rodgers, A. (2012). MEMO—a mobile phone depression prevention intervention for adolescents: development process and postprogram findings on acceptability from a randomized controlled trial. Journal of medical Internet research, 14(1), e13.
Yang, K.-M., & Cho, S.-B. (2014). Modular dynamic Bayesian network based on Markov boundary for emotion prediction in multi-sensory environment. Paper presented at the ICNC.
Yang, Y.-H., & Chen, H. H. (2011). Prediction of the distribution of perceived music emotions using discrete samples. IEEE Transactions on Audio, Speech, and Language Processing, 19(7), 2184-2196.
Zhang, Y., Tang, J., Sun, J., Chen, Y., & Rao, J. (2010). Moodcast: Emotion prediction via dynamic continuous factor graph model. Paper presented at the 2010 IEEE International Conference on Data Mining.