Machine Learning for Mental Health Screening

Hernandez-Reisch, Miranda; Caouette, Hunter C.; Bruneau, Connor D.; Kayastha, Rimsha; Melican, Veronica R.

Student Work

Machine Learning for Mental Health Screening

Public

Mental health disorders such as depression are prevalent in both the United States and the world. Left untreated, such conditions can greatly decrease the quality of one’s life and even lead to suicide. Therefore, accurate screening methods for mental health are a necessity. Surveys are commonly used but can be biased and perceived as intrusive, so there is a need for passive screening methods. This project builds on three previous years of MQP research that aimed to develop passive mental health screening methods. We made improvements to the Android and website surveys developed by previous teams. In addition, we collected two new datasets: one to investigate how students are affected by depression and another that aimed to answer remaining research questions about the mobile application survey in order to improve it. We refined the existing machine learning pipeline to increase efficiency and usability. Finally, we investigated the potential of using time series constructed from text and call logs to predict depression. Overall, this work contributed to the development of non-intrusive passive mental health screening methods that will facilitate faster diagnosis and treatment for those affected.

This report represents the work of one or more WPI undergraduate students submitted to the faculty as evidence of completion of a degree requirement. WPI routinely publishes these reports on its website without editorial or peer review.

Creator

Subject

Publisher