From PHONT Project Wiki
PGRN Ontology Network Resource (PHONT)
Our goals are to support meta-analyses, achieve translational goals, and facilitate the messaging of pharmacogenomics-related data from and potentially into clinical environments such as EMRs. Read more.
- We are analyzing data dictionaries from all PGRN sites and mapping the data elements with these data dictionaries to standardized data models and terminologies. We will be publishing the results of this analysis on our page shortly. Please email your questions to PHONT
- Ongoing collaborats with PharmGKB are taking place to incorporate terms from RxNorm and NDF-RT into their data set.
- We are collaborating with PG-POP to create standardized representations of phenotype algorithms to mine EMR data.
PGRN Data Dictionary Standardization
- PHONT is developing data standards and related value sets based on data dictionaries submitted by PGRN research sites. The proposed data standards are based on the SHARPn Clinical Element Models, which will facilitate secondary use and data integration.
- A phased approach is being used to develop proposed data standards and value sets for the PGRN. The first sets of proposed standards will be relatively simple. Those foundational data standards will be used as building blocks to represent more complex data elements, including derived data. We welcome community feedback on these proposed data standards.
SSRI Electronic Medical Record (EMR) Phenotyping
- Hypothesis: SSRI (Selective Serotonin Reuptake Inhibitor) treatment response for major depressive disorder (MDD) diagnoses is influenced by clinical risk factors in conjunction with genetic factors.
- Aim 1: To study the prevalence and association of clinical risk factors in subjects exposed to SSRIs diagnosed with MDD using EMRs.
- Aim 2: To determine if the associations vary between patients who had remission/responded to SSRIs versus who did not, and study treatment outcomes and drug side effects using EMRs.
- Methods: We have developed a preliminary electronic medical record (EMR) algorithm for identifying individuals diagnosed with MDD, and subsequently administered Celexa and/or Lexapro, and determine their treatment response outcomes (primarily PHQ9 scores) at different time intervals. EMR data was also leveraged for extracting relevant clinical risk factors and co-morbidities for the cohort.
- Preliminary Results: 140 individuals (out of 779) from the Mayo-PGRN-SSRI cohort had PHQ9 scores both at “baseline” and “1 month". On average percentage changes in PHQ9 scores in 1 month was considerably lower than average percentage changes in HAMD or QIDS scores in 4 weeks. Furthermore, remission rates by PHQ9 are higher than remission rates by HAMD/QIDS; but response rates are lower.
- Next steps: Our goal is expand the scope, and implement this algorithm within Mayo's biobank population (N=~28,000), and perform an EMR-derived treatment response pharmacogenomics SSRI study.
Pharmacogenomics Guidelines Model
- A formal model was developed to represent the CPIC guidelines. The model incorporates standard identifiers for genes, genetic variants, and drugs to facilitate integration with drug order entry and CDS systems, and to provide a mechanism to link the guidelines to public databases. The model also includes cross-references to FDA drug label information. The model is expressed in UML and converted to an XSD schema, which can be used to generate and validate XML representations of the CPIC guidelines.
- This model will facilitate adoption of the CPIC guidelines by simplifying their integration into existing clinical infrastructure. Expressing the guidelines in a computable format will streamline the expansion and maintenance of the guidelines over time, reduce the potential for human error, and help to ensure more consistent implementation of drug dosing guidelines across institutions.
- Detailed clinical models are a standardized model that serve as the basis for retaining computable meaning when data is exchanged between heterogeneous computer systems. Detailed clinical models are also the basis for shared computable meaning when data is referenced in decision support logic.
- PHONT is leveraging work on a Core set of CEMs created by a collaboration between Mayo Clinic and Intermountain Healthcare as part of the SHARPn group at Mayo Clinic.
- PHONT utilizes a wide variety of computational infrastructure to support data harmonization and standardization activities. This infrastructure includes terminology services that provide access to standard terminologies, as well as metadata repositories and harmonization platforms.
- PHONT provides access to standard terminologies that meet U.S. Meaningful Use requirements. Value sets related to pharmacogenomics data sets will be defined using these terminologies.