# The Dynamics of Mood in Bipolar Disorder: How Mathematical Models Help Phenotype Individuals, Forecast Mood, and Clarify Underlying Mechanisms

## TL;DR

This paper reviews how mathematical models can help understand and predict mood changes in bipolar disorder, offering new insights into its complex dynamics.

## Contribution

The paper systematically reviews how mathematical modeling can phenotype bipolar disorder, forecast mood, and reveal underlying mechanisms.

## Key findings

- Mathematical models can differentiate bipolar disorder from other disorders and track treatment response.
- Mood prediction using models becomes less accurate within a few days, suggesting inherent unpredictability in bipolar disorder.
- Mood states like mania and depression are best represented as regions within continuous latent dimensions.

## Abstract

Mood in bipolar disorder (BP) fluctuates in complex and unpredictable ways that resist simple explanation. To capture this complexity, researchers have turned to modeling mood dynamics. This review organizes the recent literature around three key questions: How can modeling help phenotype BP? Can models accurately predict future mood? And can modeling clarify mechanisms underlying mood instability?

Models of mood dynamics carry clinically relevant information beyond standard measures, differentiating BP from other disorders and reflecting treatment response and impairment. Yet their ability to forecast mood tends to deteriorate within days, raising the possibility that fleeting predictability may be intrinsic to BP, not simply a technical limitation. Mood also appears to be best represented by two or more continuous latent dimensions, with states like mania and depression appearing as loosely defined regions. However, how mood unfolds over time remains poorly understood, particularly the relative influence of internal dynamics versus external triggers.

Looking ahead, two priorities stand out: applying these models in real time to support clinical and research decisions, and validating them systematically through standardized forecasting tasks and shared longitudinal datasets integrating mood and physiological data. Together, these steps can help turn modeling into a practical tool for understanding and managing mood in BP.