Using a Generative Model to Characterise and Target Affective Instability in Patients with Bipolar and Borderline Personality Disorders
Affective instability predicts poor outcomes across a range of psychiatric presentations suggesting it as a potential treatment target. However, it is not clear what types of affective instability exist, how they are associated with specific diagnoses or whether they may be individually targeted by treatments. We used a generative model of affect to define two forms of affective instability, volatility and noise. We demonstrated diagnostic specificity, with volatility being raised in patients with bipolar disorder and noise in patients with borderline personality disorder and show that lithium acts specifically to increase the volatility of positive affect. These results indicate that distinct subtypes of affective instability are associated with different clinical presentations, that the subtypes can be individually targeted by interventions and suggests a novel mechanism of action for lithium.