Improving Rare Disease Classification Using Imperfect Knowledge Graph

Accurately recognizing rare diseases based on symptom description is a critical task. The lack of historical data for rare diseases poses a great challenge to machine learning-based approaches. In this study, we develop a text classification algorithm that represents a document as a combination of a “bag of words” and a “bag of knowledge terms where a “knowledge term” is a term shared between the document and the knowledge graph relevant to the disease classification task.