Data Science | Ph.D. Qualifying Exam | Monica Tlachac | “Predicting PHQ-9 Depression Scores with Smartphone Text Messages.”

Wednesday, January 09, 2019
3:00 pm
Floor/Room #: 
Beckett Conference Room, 2nd Floor


Ph.D. Research Qualifying Exam
Monica Lauren Tlachac

January 9, 2019 | 3:00 pm
Beckett Conference Room
Fuller Laboratories, 2nd Floor

Advisor: Prof. Elke Rundensteiner, Ph.D.

Prof. Eleanor Loiacono, Ph.D.

Prof. Mohamed Eltabakh, Ph.D.



“Predicting PHQ-9 Depression Scores with Smartphone Text Messages.”

Depression, a serious and debilitating mental illness, is frequently undiagnosed.  This can be due to not recognizing symptoms, lack of access to medical resources, or fear of stigma.  However, diagnosis is important as it is the first step towards treatment.  The PHQ9 is a 9-question screening tool for depression.  Many other studies have used a variety of data types and features to predict the PHQ-9 scores for individuals.  In this study, we focus on the predictive power of a single underutilized modality: received text messages.  We extract polarity and subjectivity features from text messages collected from 315 participants over two-weeks for different subsets of contacts.  We demonstrate that machine learning methods can exploit the received texts of to predict an individual’s PHQ-9 score with high F1 score.