Using a longitudinal model to predict outcomes to ketamine intravenous therapy for depression

Ketamine Intravenous Therapy (KIT) is an effective, rapid-acting treatment for depression, but individual patients’ responses may be highly variable over time. This natural variation in a patient's mood state at any given time point complicates efforts to predict response trajectories. One way to overcome variability in response within an individual is to characterize patient response based on longitudinal outcomes rather than a single end-point. Prior studies have tackled this challenge by classifying longitudinal responses into a few distinct patterns of response to KIT1,2.

Here, we take an alternative approach: to model individual responses to KIT as individual variations from the “average” response, rather than as distinct categories of responders. Using a large sample of patients (N = 4991) from the Osmind Real-World Database. We build a highly detailed statistical model of the population and individual longitudinal response to KIT. We use patient-related clinical features to predict the individual longitudinal response to KIT.