A Personalized Support Agent for Depressed Patients: Forecasting Patient Behavior Using a Mood and Coping Model

Depression is a disorder that has a huge impact on both the patient and its environment. An effective treatment of depression is of crucial importance. Currently, Internet-based self-help therapies are the state-of-the-art among therapies that do not involve a human therapist. However, these interventions are not tailored towards individual patient needs. The utilization of pervasive technology, including a mobile phone and its sensors could potentially provide a way to make therapies more personalized and accessible at any time. One crucial aspect to make such personalization possible is to understand the current state of the patient and the ability to make a prediction on the expected state of the patient in the future. Obviously, predictions can differ greatly per patient. This paper takes a cognitive modeling approach in which the parameters of the model can be adapted to the characteristics of the patient. Hereby, an existing model for mood and coping is taken as a basis and different techniques are proposed to tailor the model towards the patient using sensory information that has been obtained. An evaluation is performed using a dataset from the psychological domain.