Grant: Project Grant

What do we want to do? The overall objective of this study is to characterize illness trajectories and to predict relapses in stable adult patients with Bipolar Disorder (BD). Specifically, we aim to develop new data modeling that will enable more tailored signal detection to examine changes within each individual over time, to predict a future episode of illness. Our preliminary studies relevant to this proposal include those related to using nonlinear techniques (time-series analysis and entropy) to understand how mood is regulated in healthy controls, bipolar patients and their relatives; as well as our ongoing study analyzing heart rate variability and clinical characteristics in BD patients during different phases of their illness (stable, depressed, hypomanic or mixed). How are we going to do it? We will collect data using passive sensing (Oura ring), including sleep, activity and heart rate variability, in 200 BD patients. This ring measures these variables continuously, so we will have a large amount of data. These data will be combined with a detailed description of the clinical course of illness in each patient. Then, we will use nonlinear techniques to obtain an individualized measure of variability (entropy levels) in each participant. Finally, we will feed these entropy levels for each participant into a machine-learning technique called Markov Brains to distinguish those at a higher risk of relapse. Why is this important? Because BD is a complex and heterogeneous illness. This means that we cannot assume that all depressive episodes or all manic episodes will look the same in all patients diagnosed with BD. Thus, we need to incorporate each patient's variability (e.g., the unique way in which each patient's mood or sleep change) to extract individual patterns suggestive of a future episode of illness. The long-term benefits of our studies include reducing rates of relapse and related complications, including substance abuse and suicide.