Predicting Mood Changes in Bipolar Disorder Through Heartbeat Nonlinear Dynamics

Bipolar disorder (BD) is characterized by an alternation of mood states from depression to (hypo)mania. Mixed states, i.e., a combination of depression and mania symptoms at the same time, can also be present. The diagnosis of this disorder in the current clinical practice is based only on subjective interviews and questionnaires, while no reliable objective psycho-physiological markers are available. Furthermore, there are no biological markers predicting BD outcomes, or providing information about the future clinical course of the phenomenon. To overcome this limitation, here we propose a methodology predicting mood changes in BD using heartbeat nonlinear dynamics exclusively, derived from the ECG. Mood changes are here intended as transitioning between two mental states: euthymic state (EUT), i.e., the good affective balance, and non-euthymic (non-EUT) states. Heart rate variability (HRV) series from 14 bipolar spectrum patients (age: 33.43 ± 9.76, age range: 23-54; six females) involved in the European project PSYCHE, undergoing whole night electrocardiogram (ECG) monitoring were analyzed. Data were gathered from a wearable system comprised of a comfortable t-shirt with integrated fabric electrodes and sensors able to acquire ECGs. Each patient was monitored twice a week, for 14 weeks, being able to perform normal (unstructured) activities. From each acquisition, the longest artifact-free segment of heartbeat dynamics was selected for further analyses. Sub-segments of 5 min of this segment were used to estimate trends of HRV linear and nonlinear dynamics. Considering data from a current observation at day t 0 , and past observations at days (t -1 , t -2 ,...,), personalized prediction accuracies in forecasting a mood state (EUT/non-EUT) at day t +1 were 69% on average, reaching values as high as 83.3%. This approach opens to the possibility of predicting mood states in bipolar patients through heartbeat nonlinear dynamics exclusively.