Facilitating early detection of depression through conversational audios and machine learning techniques

Abstract

Mental health is becoming a trending topic amongst society. The relevance of it in our lives is being studied in order to achieve a better comprehension for our well-being. Studies have shown that both anxiety and depression greatly affect higher education student’s performance and development, as well as post-graduate life. Early detection of depression, or other mental health issues, could lead to sooner evaluation and support. As humans go through life, many stressful situations arise. This is not possible to avoid. Nevertheless, our resilience to stress is the factor that estimates how much stress we can handle until reaching alerting levels of a possible mental disorder. This research intends to use machine learning techniques to deliver an accurate classification from depressive indicators based on conversational audios. The result provided will be used by an algorithm to analyze the individual’s state, and with the combination of conversational audios and the psychophysiological profile, it will identify early symptoms of the illness, which will alert the individual in time to act.