Enhancing random forest classification with NLP in DAMEH: A system for DAta Management in eHealth Domain

摘要: Abstract The use of pervasive IoT devices in Smart Cities, have increased the Volume data produced many and field. Interesting very useful applications grow up number E-health domain, where smart are used order to manage huge amount data, highly distributed environments, provide services able collect fill medical records patients. problem here is gather produce analyze depending on their contents. Since gathering involve different (not only wearable sensors, but also environmental devices, like weather, pollution other sensors) it difficult classify contents, enable better management Data from couple with written natural language: we describe an architecture that determine best features for classification, existent records. based pre-filtering phase Natural Language Processing, enhance Machine learning classification Random Forests. We carried experiments about 5000 real (anonymized) case studies various health-care organizations Italy. show accuracy presented approach terms Accuracy-Rejection curves.