Semi-automatic Mapping of Medical Data onto Ontologies

Harvard Center for Computational Biomedicine Mathematics, 2021–22

Liaison(s): Dr. Rafael Goncalves, Dr. Robert Gentleman
Advisor(s): Jamie Haddock
Students(s): Carmen Benitez, Cindy Lay (PM-F), An Nguyen, Kobe Rico (PM-S), Matthew Waddell

Harvard’s Center for Computational Biomedicine designs methods and tools to aid scientists to discover, integrate, and carry out meta-analyses over heterogeneous sources of data, such as using ontologies to effectively structure and search data. The Harvard CCB Clinic Team is building upon an existing prototype tool for semi-automatically mapping phenotype terms to controlled terms in ontologies in three ways. First, they are implementing graphical user interfaces to use the tool and to interactively browse and verify the automatic mappings. Second, they are testing and implementing NLP-based, semantic methods to improve the tool’s mapping performance. Third, they are creating an automated testing harness to measure the tool’s mapping performance as changes are made.