A Framework for Identifying Genotypic Information from Clinical Records: Exploiting Integrated Ontology Structures to Transfer Annotations Between ICD Codes and Gene Ontologies

Seyedsasan Hashemikhabir,Ran Xia,Yang Xiang,Sarath Chandra Janga

DOI: https://doi.org/10.1109/tcbb.2015.2480056

2018-01-01

IEEE/ACM Transactions on Computational Biology and Bioinformatics

Abstract:Although some methods are proposed for automatic ontology generation, none of them address the issue of integrating large-scale heterogeneous biomedical ontologies. We propose a novel approach for integrating various types of ontologies efficiently and apply it to integrate International Classification of Diseases, Ninth Revision, Clinical Modification (ICD9CM), and Gene Ontologies. This approach is one of the early attempts to quantify the associations among clinical terms (e.g., ICD9 codes) based on their corresponding genomic relationships. We reconstructed a merged tree for a partial set of GO and ICD9 codes and measured the performance of this tree in terms of associations' relevance by comparing them with two well-known disease-gene datasets (i.e., MalaCards and Disease Ontology). Furthermore, we compared the genomic-based ICD9 associations to temporal relationships between them from electronic health records. Our analysis shows promising associations supported by both comparisons suggesting a high reliability. We also manually analyzed several significant associations and found promising support from literature.