Screening for bipolar disorder in a tertiary mental health centre using EarlyDetect: A machine learning-based pilot study

Abstract

Background: : Bipolar disorder (BD) is a prevalent mental health illness with a direct impact on patient's well-being. Self-report-based BD screening questionnaires such as the Mood Disorder Questionnaire (MDQ) is economical and clinically validated. We use a machine-learning approach to test whether utilizing our composite screening application - EarlyDetect (ED), designed for assessing an array of mental health illness, can enhance bipolar disorder screening over MDQ. Methods: : This was a retrospective, naturalistic study at a tertiary mental health centre in western Canada. Participants (n = 955; 56.4% female; mean age 35.4; 18.7% BD) completed ED and underwent a clinical interview with a blinded psychiatrist for diagnostic accuracy. Elastic net and leave-one-out cross-validation was used to make more confident predictions at an individual level. Results: : Using composite scoring, the balanced accuracy of our tool was 80.6%, with a sensitivity of 73.7% and a specificity of 87.5%. Compared with the MDQ original scoring method, the fully composite ED model improved balanced accuracy by 6.9%, sensitivity by 14.5%, while maintaining specificity. Limitations: : Patients were assessed using clinical psychiatric evaluations, which are subjective. There is also the potential for self-reporting bias. BD subtypes were not differentiated. The cross-sectional design of this study rules out conclusions of causality. Conclusions: : Our results show improved BD detection accuracy using composite measures.

Keywords

• Bipolar screening
• Bipolar disorder
• Machine-learning
• Mobile app
• m-health