Enriching Representation Learning Using 53 Million Patient Notes through Human Phenotype Ontology Embedding

Abstract

The Human Phenotype Ontology (HPO) is a dictionary of more than 15,000 clinical phenotypic terms with defined semantic relationships, developed to standardize their representation for phenotypic analysis. Over the last decade, the HPO has been used to accelerate the implementation of precision medicine into clinical practice. In addition, recent research in representation learning, specifically in graph embedding, has led to notable progress in automated prediction via learned features. Here, we present a novel approach to phenotype representation by incorporating phenotypic frequencies based on 53 million full-text health care notes from more than 1.5 million individuals. We demonstrate the efficacy of our proposed phenotype embedding technique by comparing our work to existing phenotypic similarity-measuring methods. Using phenotype frequencies in our embedding technique, we are able to identify phenotypic similarities that surpass the current computational models. In addition, we show that our embedding technique aligns with domain experts judgment at a level that exceeds their agreement. We show that our proposed technique efficiently represents complex and multidimensional phenotypes in HPO format, which can then be used as input for various downstream tasks that require deep phenotyping, including patient similarity analyses and disease trajectory prediction.