Browsing by Author "Green, Julie Meadows"

Now showing 1 - 2 of 2
  • Thumbnail Image
    Considerations for Automating Salmonella Serovar Identification within an Electronic Public Health Reporting Environment
    Alexander, Jeffry Chanen (Virginia Tech, 2015-09-08)
    CDC's requirements for Salmonella surveillance reporting include submission of serovars from the recognized naming scheme, Kauffmann-White (K-W), using identifiers curated by the Systematized Nomenclature of Medicine-Clinical Terms (SNOMED CT®). Translating the serotype formula of a Salmonella isolate to the correct identifier has been a multistep manual process for users. Our goal was to determine whether a degree of automation could be achieved using an ontology based on K-W. We investigated information artifacts presently available, namely K-W, SNOMED CT and CDC's Public Health Information Network - Vocabulary Access and Distribution System (PHIN-VADS). As SNOMED CT creates identifiers and associates them with serovar names, we performed detailed analysis on its coverage of K-W. An overall error rate of 13.1% included simple omissions and transcription errors. We limited our assessment of K-W and PHIN-VADS to the functional characteristics of the resources they distribute. K-W creates serovar names but does not provide identifiers. PHIN-VADS includes the identifiers but not antigenic formulae for most isolates. In summary, neither K-W nor PHIN-VADS contained all information users require. Two different ontology prototypes were developed. Prototype I placed K-W serovars as terminal nodes in the hierarchy and these were given logic-based definitions. Prototype II added isolate classes as serovar subtypes. Only the isolate classes had complete logical definitions. Both prototypes were logically sound and functioned as expected. Prototype I paralleled existing SNOMED CT content but required more robust description logic than currently employed in SNOMED CT. Prototype II was more compatible with current functionality of SNOMED CT but created identifiers that would not meet current requirements for public health reporting. Prototype I was fully populated as the Salmonella Serotype Designation Ontology (SSDO). As it stands, SSDO reliably places isolates in the appropriate classes, with few and predictable exceptions. Although SNOMED CT cannot accommodate its functionality at this time, SSDO can serve as the basis for a stand-alone application. Ultimately whether by improving functionality of existing systems or providing a framework for an ancillary automated system, this work should facilitate real-time reporting and analysis of surveillance data that will prevent new or reduce severity of infectious disease outbreaks.
  • Thumbnail Image
    Development and evaluation of methods for structured recording of heart murmur findings using SNOMED CT® post-coordination
    Green, Julie Meadows (Virginia Tech, 2004-12-10)
    Objective: Structured recording of examination findings, such as heart murmurs, is important for effective retrieval and analysis of data. Our study proposes two models for post-coordinating murmur findings and evaluates their ability to record murmurs found in clinical records. Methods: Two models were proposed for post-coordinating murmur findings: the Concept-dependent Attributes model and the Interprets/Has interpretation model. A micro-nomenclature was created based on each model by using the subset and extension mechanisms provided for by the SNOMED-CT® framework. Within each micro-nomenclature a partonomy of cardiac cycle timing values was generated. In order for each model to be capable of representing clinical data, a mechanism for handling range values was developed. One hundred murmurs taken from clinical records were entered into two systems that were built based on each model to enter and display murmur data. Results: Both models were able to record all 100 murmur findings; both required the addition of the same number of concepts into their respective micro-nomenclatures. However, the Interprets/Has interpretation model required twice the storage space for recording murmurs. Conclusion: We found little difference in the requirements for implementation of either model. In fact, data stored using these models could be easily inter-converted. This will allow system developers to choose a model based on their own preferences. If at a later date a method is chosen for modeling within SNOMED-CT, the data can be converted to conform if necessary.