Predicting Mood Changes in Bipolar Disorder through Heartbeat Nonlinear Dynamics.

Area of Science:

• Cardiology
• Psychiatry
• Biomedical Engineering

Background:

• Bipolar disorder (BD) diagnosis relies on subjective assessments, lacking objective psychophysiological markers.
• Current methods do not predict BD outcomes or future clinical course effectively.
• There is a need for reliable, objective tools to monitor mood states in bipolar disorder.

Purpose of the Study:

• To develop a methodology for predicting mood state transitions in bipolar disorder (BD).
• To utilize exclusively nonlinear heartbeat dynamics derived from electrocardiogram (ECG) for mood prediction.
• To differentiate between euthymic (EUT) and non-euthymic (non-EUT) states in BD patients.

Main Methods:

• Analyzed Heart Rate Variability (HRV) from 14 bipolar spectrum patients using wearable ECG monitoring.
• Collected data over 14 weeks, allowing normal patient activities.
• Estimated HRV linear and nonlinear dynamics from 5-minute sub-segments of artifact-free heartbeat data.

Main Results:

• Achieved personalized prediction accuracies of 69% on average for forecasting mood states (EUT/non-EUT) 24 hours in advance.
• Reached prediction accuracies as high as 83.3% for mood state transitions.
• Demonstrated the potential of heartbeat nonlinear dynamics for predicting mood changes in BD.

Conclusions:

• Heartbeat nonlinear dynamics offer a promising, objective approach for predicting mood states in bipolar disorder.
• This methodology could enhance the monitoring and management of bipolar disorder.
• Future research can explore integrating these findings into clinical practice for personalized BD care.