In many health care institutions, caregivers record information about disease and treatment of patients in an electronic health record. Besides their importance for good healthcare, we may be able to use these for obtaining new knowledge as well. This is necessary, because our medical knowledge is far from complete, and there is a simultaneous need to keep health care affordable. Electronic health records can receive a kind of new use, by letting computers look for patterns and associations.
In this research we focus on analyzing health records in psychiatry, which focuses on mental health care. First we look at some technical, ethical and organizational requirements to start using such data. For example, how can caregivers collaborate with computer scientists, and how can data be properly anonymized?
After this we apply new techniques from artificial intelligence to data from electronic health records. We investigate whether we can learn to better assess the risk of violent behavior from patients, and whether there exist new or different diagnostic groups of patients than we know so far. The results of this research show that learning from electronic health records is a promising and feasible new approach.