Unspecific back pain is one of the most burdensome and expensive health conditions people and economies are facing worldwide. It represents a condition where no clear underlying pathology can be identified. Thus, disease mechanisms, complicating medical conditions, and differentiation in its treatment are subject to a more in-depth investigation. This project aims at identifying new data-driven subtypes and at learning about the treatment processes of patients with unspecific back pain. To that end, a large database of electronic health records is used to perform phenotyping, process mining and clustering on a back pain-specific subset of these data. A focus of the evaluation is to identify patterns that lead to unfavorable outcomes such as opioid usage, frequent recurrence and chronification of the symptoms.