Development and feasibility testing of a continuous moodsensor smartphone application to construct an early-relapse recognition model for individuals with bipolar disorder

Introduction: Using smart technology for patients with bipolar disorder could improve their accessibility and adherence to treatments and overcome the difficulties to momentary assessments in early detection of relapse signs. The present study inventively used smartphone application (APP) to construct an early-relapse recognition model (ERRM) for individuals with bipolar disorder.

Method: The ERRM APP has been developed through interdisciplinary collaboration. Forty-three patients with bipolar disorder were recruited from a medical center in Southern Taiwan. They were prospectively and longitudinally followed with the ERRM APP for 3 months. Data included passive continuous indices and self-reported mood assessments were collected. Patients’ mood symptoms were validated via clinicians’ monthly mood assessments. Deep learning techniques of artificial intelligence have been used to analyze the data of mood digital phenotype from patients’ smartphones. Additionally, 10 participants were interviewed to provide their subjective experience and responses of using this smartphone APP assessment.

Results: The ERRM APP has been supported its feasibility for individuals with bipolar disorder and the predictive model for early relapse recognition was identified.

Conclusion: The smartphone ERRM APP in Chinese is promisingly developed and tested to early recognize warning sign and mood symptoms among individuals with bipolar disorder in Taiwan. To our knowledge, this is the first mood-detection APP developed in Chinese and from research-testing. However, the small sample size may limit our ability to make generalized conclusions. Future implication will be discussed.