The Computational Psychiatry of Major Depressive Disorder

Depression is the leading cause of disability worldwide, affecting more than 300 million people. The social and economic costs of depression are enormous. Unfortunately, current antidepressant treatments do not help many of those who suffer from depression. This project proposes new approaches to assessing depression and identifying which treatments are most likely to be helpful.

It is now accepted that depression can result from a variety of different sources, much like a cough can have many different underlying causes. There is at present no reliable way for a psychiatrist to know which treatment is likely to be most effective for helping a particular depressed individual. Furthermore, researchers have not yet managed to provide a clear picture of what determines if, and when, an individual's mood will worsen and what happens in the brain when mood changes. This lack of understanding of the determinants of mood also makes it difficult to develop new treatments for mood disorders like depression. Our recent research has shown that it possible to measure momentary subjective states like happiness and that we can predict precisely how happiness will change from moment to moment during a decision-making game played on smartphones by over 18,000 players worldwide.

In this project, we will quantify how mood is determined in multiple decision-making environments. We will then ask whether this improved understanding of mood can be used to better understand depression by having healthy and depressed individuals engage in our tasks, presented in the form of games and played either in the lab or at home on smartphones.

The project has three major goals:

1) To increase knowledge of the neural circuits that determine mood in both healthy and depressed individuals.

2) To develop a new tool that uses smartphones to remotely assess depressed individuals and allows behaviour and mood data from a variety of tasks to be collected that could help clinicians make better treatment decisions.

3) To determine how different antidepressant drugs affect behaviour and mood, results that will help to understand when each drug might be most effective in treating depression.

The three 'games' developed in the project will provide measures that relate to the current state of a player's brain. For example, the games might detect that over several months an individual is becoming more and more likely to take risks, or is increasingly upset when those risks do not pay off. The numbers measured from the games provide a snapshot of the individual's current state, since they provide information about how the individual makes decisions and responds to decision outcomes that in turn reflect the workings of neural circuits affected in depression. By examining whether results of the games relate to treatment efficacy, we might be able to predict which treatment will be most effective for helping a depressed individual.

When a clinician is evaluating her patient, she might someday refer to an analysis of the patient's game scores in addition to responses to questions asked by the smartphone app. Two different depressed individuals may both have low mood but for very different reasons. The scores can in principle be used to suggest that a certain course of treatment is likely to be most effective. For example, a combination of an antidepressant medication and a specific cognitive behavioural therapy may often be effective in people with a certain set of scores that reflect the workings of neural circuits that can be affected in depression. The clinician could then use that information, in combination with her expert evaluation and her knowledge of the patient's circumstances, to make a better treatment decision. In this way, the project will, if successful, demonstrate a new way to gather rich quantitative and clinically relevant data that can complement existing clinical information and improve the treatment of depression.

Aims. I will develop tasks to quantify (using computational models) behaviour and mood in multiple decision domains affected by depression. These tasks will allow identification of candidate behavioural biomarkers for patient stratification for use in rapid assessment and treatment assignment.

Methodology. I will use computational modelling to quantify decision making and mood in three behavioural tasks, providing a wide range of measures that might be affected by antidepressant medication and by depression. Functional MRI scanning of depressed patients will allow examination of the neural circuits that determine mood and behaviour in health and in depression. I will use pharmacological studies in healthy and depressed subjects to examine effects of two different types of antidepressants (a serotonergic drug and a glutamatergic drug) on behaviour and mood. I will combine lab testing with smartphone-based data collection to longitudinally track remitted medication-free patients over 14 months, to look for predictors of relapse based on behaviour/mood model parameters.

Scientific and Medical Opportunities. This project will provide important insights into the computational and neurobiological mechanisms that determine mood in healthy and depressed subjects. It will test a novel lab-smartphone approach that aims to provide longitudinal data in depressed subjects that may predict relapse probability. Behaviour/mood parameter measurements before ketamine treatment may be useful for predicting which patients are mostly likely to respond to that treatment. Pharmacological results will increase knowledge of the roles of serotonin and glutamate in decision making and mood. Clearer understanding of the neural mechanisms that underlie decision making and mood that are affected by depression and antidepressant treatment could have significant medical benefit for improving the assessment and treatment of depression.

This work will impact:

1) Individuals suffering from depression - by improving our understanding of the neural circuitry underlying depression, and testing a new tool that combines smartphone tasks with models of behaviour/mood that may improve patient assessment and treatment development on the time scale of 7-10 years.

2) Clinicians - by demonstrating the utility of an approach that can provide longitudinal assessment of patients that complements traditional clinical measures and helps clinicians makes treatment decisions, an approach that might be employed on the time scale of 7-10 years.

3) Staff on the project - staff trained by Dr Rutledge including the postdoctoral researcher, research assistant, and MSc students who assist research on the project will receive training in using quantitative methods to address neuroscience and mental health issues, skills which are of widespread use in academia, industry, and the healthcare sector.

4) Business - the approach has potential value for businesses including pharmaceutical companies because it might permit novel antidepressant treatments to be tested more quickly than is currently possible and such an approach could be employed on the time scale of 7-10 years.

5) Health care providers - an inexpensive and effective tool for remotely assessing and monitoring depression treatment would lower healthcare costs by reducing the use of ineffective treatments and the frequency of doctor appointments needed to effectively treat patients on the time scale of 10-15 years.

6) General public - by increasing discussion of mental health issues that reduce the stigma of discussing depression and helping the families of depressed individuals understand what depression is. I will use a combination of blogs/websites/social media and more novel approaches as I have in the past including theatre (for an art-science project on happiness with the Roundhouse theatre) and smartphones so that the public can contribute to and learn about my research.

7) UK society and economy - any significant improvements in the assessment or treatment of depression would have societal benefits in terms of improved health and quality of life and corresponding economic benefits due to the increased workplace productivity of more effectively treated patients on the time scale of 10-15 years.