Exploring Rare Disease Similarities and Matching

Abstract

In numerous contexts, similarity serves as a way to determine if several entities are part of the same category or display common characteristics. Likewise, matching is a broad concept that involves the ability to align one entity with another. Both matching and similarity are essential in the Semantic Web, particularly for searching and classification. The biomedical sector has emerged as a particularly dynamic area for applying the Semantic Web, where these functions are critical. Matching techniques aid in differential diagnosis, while similarity searches aid in classifying diseases. Recently, there has been an increased emphasis on diagnosing and categorizing rare diseases within this framework. This paper emphasizes the phenotypic information of rare diseases and formally defines certain similarity and matching relations among rare diseases, including matching ratios and similarity scores. The proposed relations and metrics are designed to advance the current rare disease classification methods and matching procedures by considering the frequency of phenotypes and their hierarchical relationships.