From Music to Emotions and Tinnitus Treatment, Initial Study

Abstract

The extended tinnitus database consisting of 758 patients with information repeated from the original database of 555 patients, along with the addition of visits and a new questionnaire, the Tinnitus Function Index and Emotion Indexing Questionnaire, is used to mine for knowledge. New patients in the extended database represent those patients that have completed the Tinnitus Function Index questionnaire (TFI) [10]. The patient visits are separated and used for mining and action rule discovery based on all features and treatment success indicators including several new features tied to emotions (based on a mapping from TFI to Emotion Indexing Questionnaire (EIQ) [14]; EIQ questionnaire is used by our team to build personalized classifiers for automatic indexing of music by emotions). We propose a link between TFI and EIQ leading to a creation of new features in the extended tinnitus database. Then, we extract knowledge from this new database in the form of association action rules to assist with understanding and validation of diagnosis and treatment outcomes.