Spelta-Miner: An Expert System Based On Data Mining And Multilabel Classification To Design Therapy Plans For Communication Disorders

The ability to express feelings and needs, to share ideas or to establish relations is fundamental in different stages of human life. During language acquisition, a child develops several intellectual, psychological and emotional skills that support many other brain functions. On those grounds, speech-language therapy is a fundamental process for those people who suffer from communication disorders. However, currently in several developing countries there is a lack of rehabilitation services, and personnel of many centers overwork. In this paper, we present an expert system able to design therapy plans for patients suffering from disabilities related with communication. The system uses an approach based on natural language processing and multi-label classification to determine which strategies must be carried out in a therapy session. This approach has been tested on a database consisting on 1,345 strategies and 53 real cases of children with disabilities. In order to evaluate the generated plans we have measured accuracy and quality, obtaining encouraging results.