Chronic fatigue syndrome (CFS) is a debilitating illness characterised by extreme tiredness, with a wide range of related symptoms, including muscle fatigue and sleep disturbance [1]. Given that the origin of the condition is still unknown, analysing large amounts of patient data is an important step towards an improved understanding of its aetiology and clinical trajectories. Clinical assessments of CFS provide a comprehensive overview of patients’ symptoms and functional limitations, and are usually conducted when a patient is referred to a specialist clinic. These assessments include past history as well as current situation, thus representing a valuable source of data for large-scale analysis. In electronic health records (EHRs), however, clinical assessments are often primarily documented in an unstructured form (free text), which cannot be easily analysed on a large scale. To allow the automatic analysis of such texts, natural language processing (NLP) methods are becoming increasingly popular [2].