Amanda Hicks, Ph.D., received a $19,910 grant from the UF CTSI Pilot Award program to create Hypertension FACTS, a tool that would aid in computer retrieval and analysis of data on the condition.
Hicks, an assistant professor in the Department of Health Outcomes & Policy, says Hypertension FACTS (Florida Annotated Corpus for Translational Science) will consist of a group of existing published case reports that have been annotated with ontologies. This will help clinicians and researchers more easily find, retrieve and use information about hypertension stored in electronic health records and the medical literature.
Hicks, who is also a member of the biomedical informatics team with UF’s CTSI, defines ontologies as “formal representations of any of the following: a) things in the world, b) knowledge and concepts relevant to a particular domain, or c) the meanings of terms and data.”
“They [ontologies] contain definitions that aid both human understanding and computational parsing of meaning,” she said. “As machine-readable representations of knowledge, biomedical ontologies render biomedical knowledge computable, thereby supporting computer-aided discovery of knowledge that would otherwise lie dormant in databases and scientific text.”
“Ontologies are crucial tools for data integration and knowledge discovery, especially as the wide variety of existing data takes on increasing importance as a resource for researchers and clinicians,” Hicks said.
Hicks plans to develop a manually annotated corpus of 20 open-access, full-text PubMed case reports about hypertension using the Open Biomedical Ontologies (OBO) Library’s ontologies of diseases, symptoms, vital signs, diagnostic procedures, and drugs.
“The project takes a novel and significant step toward integrating patient-level data with disease and drug classification systems,” she said. “The result will be a freely available gold standard corpus of case reports about hypertension, along with supporting resources that researchers can use to develop similar gold standards in other health care domains.”
Hicks’ focus on hypertension also will contribute to a key research area of the OneFlorida Clinical Research Consortium’s clinical data research network (CDRN). The consortium includes three large university systems and nine clinical systems that provide care for nearly half of all Floridians through 4,100 physicians, 1,240 clinical practices, and 22 hospitals. Its centralized infrastructure includes a central Data Trust containing de-identified electronic health records for some 10.2 million Floridians. The consortium is one of 13 national CDRNs affiliated with the Patient-Centered Outcomes Research Institute in Washington, D.C. Each CDRN in the network dedicates a portion of its research efforts toward a prevalent health condition. The OneFlorida Clinical Research Consortium’s prevalent clinical health condition is hypertension.
Hicks said practical applications for the Hypertension FACTS corpus include training a computer-based system to retrieve patient data from written text and developing data-retrieval systems to match clinical phenotypes (data profiles) of patients with those described in case reports.