The rate of mental health disorders is rising across the globe. While it significantly affects the quality of life, an early detection can prevent the fatal consequences. Existing literature suggests that mobile based sensing technology can be used to determine different mental health conditions like stress, bipolar disorder. In today's smartphone based communication, a significant portion is based on instant messaging apps like WhatsApp; thus providing the opportunity to unobtrusively monitor the text input interaction pattern to track mental state. We, in this paper, leverage on the text entry pattern to track multiple emotion states. We design, develop and implement an Android based smartphone keyboard EmoKey, which monitors user's typing pattern and determines four emotion states (happy, sad, stressed, relaxed) by developing an on-device, personalized machine learning model. We evaluate EmoKey with 22 participants in a 3-week in-the-wild study, which reveals that it can detect the emotions with an average accuracy (AUCROC) of 78%.
International Conference on Communication Systems and Networks
Distributed and Interactive Systems

Ghosh, S., Sahu, S., Ganguly, N., Mitra, B., & De, P. (2019). EmoKey: An emotion-aware smartphone keyboard for mental health monitoring. In International Conference on Communication Systems and Networks. doi:10.1109/COMSNETS.2019.8711078