Ultra-brief questionnaires for pre-screening for depressive symptomatology
Research suggests the first two items of the Patient Health Questionnaire-9 (PHQ-9), the PHQ-2, are the cardinal symptoms of depression and are therefore the best questions for identifying depressive symptomatology. The PHQ-2 has shown strong identification performance when compared with other longer questionnaires. However, it has not been compared with other item pairings within the PHQ-9. Therefore, individuals may be incorrectly classified, and their depressive symptoms missed if sub-optimal ultra-brief questionnaires are used in screening interviews.
The objective of the project is to use machine learning to optimise currently used questionnaires for the detection of mental disorders. Machine Learning provides a data-driven approach to psychology, giving new insights into the symptoms of depression that may have been overlooked by the current consensus-based approach of psychology. By applying machine learning methods to existing datasets, this project seeks to reduce the size of questionnaires whilst retaining their predictive power. At scale, the use of these shortened questionnaires can reduce screening time by only prompting the completion of a full questionnaire or a psychiatric assessment if probable depression is detected.