[[SSRN-id4045955.pdf]] [[DP - Sleep]] #mood_regulation #EMA - Aim: To predict impending mood episodes recurrences using digital phenotypes related to circadian rhythm obtained from wearable devices and smartphones - Observational Study. Young patients, with MDD, BD 1, BD 2. Total 495 patients from 8 hospital in South Korea. - Followed up average 279.7 days. - Algorithm predicted mood episodes were compared to those identified through face-to-face clinical interviews. - Outcome: ==The prediction accuracies for impending major depressive episodes, manic episodes and hypomanic episodes for the next three days were 90.1%, 92.6% and 93% with the area under the curve values of 0.937, 0.957, and 0.963, respectively== - Currently, we relay too much on patients' self-reported symptoms. ## What is the link between Mood and Sleep? --- Using circadian rhythm multi-centre, observational study of young adults predict impending mood episodes, compared with face-to-face clinical interviews. (every 3 months) 90% accuracies The shift of circadian rhythm precedes worsening of mood symptoms. Sleep-wake problems were evident even during euthymic periods of mood disorders. Sleep pattern is important, and medication is considered to be mediating circadian. (11) ## methods - Recruited participants - - then use EMA (eMoodChart) to collect mood and energy. - Likert scale (-3 to 3) - Wear Fitbit - Clinical Assessment, every 3 months interview. - Clinicians blinded - Passive tracking - Ambient lighting, Steps, HR, Sleep. - They use light exposure, step, sleep, and circadian rhythm - as features that can be associated with circadian rhythm disruption. # Prediction Model Construction - Random Forest - Supervised Learning Algorithm to train the model. - Major Depressive Episode - Manic Episode - Hypomanic episode - Retrospectively, the algorithm looks at collected data (18 days before the established episode) - " a set of optimal parameters were deterred and used to predict whether a mood episode occurred in the next 3 days" - Then they compare with EMA and Self-report symptoms. Question: How do they know the patient has all these mood episodes? ------ - Do patients in Korea have different presentations compared to Singapore? - We probably need a baseline of Singapore patients' circadian baseline. - What does a "normal", "healthy" person's circadian baseline look like? how will that look compared to a patient?