Experiments in concept modeling for radiographic image reports.

OBJECTIVE: Development of methods for building concept models to support structured data entry and image retrieval in chest radiography. DESIGN: An organizing model for chest-radiographic reporting was built by analyzing manually a set of natural-language chest-radiograph reports. During model building, clinician-informaticians judged alternative conceptual structures according to four criteria: content of clinically relevant detail, provision for semantic constraints, provision for canonical forms, and simplicity. The organizing model was applied in representing three sample reports in their entirety. To explore the potential for automatic model discovery, the representation of one sample report was compared with the noun phrases derived from the same report by the CLARIT natural-language processing system. RESULTS: The organizing model for chest-radiographic reporting consists of 62 concept types and 17 relations, arranged in an inheritance network. The broadest types in the model include finding, anatomic locus, procedure, attribute, and status. Diagnoses are modeled as a subtype of finding. Representing three sample reports in their entirety added 79 narrower concept types. Some CLARIT noun phrases suggested valid associations among subtypes of finding, status, and anatomic locus. CONCLUSIONS: A manual modeling process utilizing explicitly stated criteria for making modeling decisions produced an organizing model that showed consistency in early testing. A combination of top-down and bottom-up modeling was required. Natural-language processing may inform model building, but algorithms that would replace manual modeling were not discovered. Further progress in modeling will require methods for objective model evaluation and tools for formalizing the model-building process.

The Canon Group's effort: working toward a merged model.

OBJECTIVE: To develop a representational schema for clinical data for use in exchanging data and applications, using a collaborative approach. DESIGN: Representational models for clinical radiology were independently developed manually by several Canon Group members who had diverse application interests, using sample reports. These models were merged into one common model through an iterative process by means of workshops, meetings, and electronic mail. RESULTS: A core merged model for radiologic findings present in a set of reports that subsumed the models that were developed independently. CONCLUSIONS: The Canon Group's modeling effort focused on a collaborative approach to developing a representational schema for clinical concepts, using chest radiography reports as the initial experiment. This effort resulted in a core model that represents a consensus. Further efforts in modeling will extend the representational coverage and will also address issues such as scalability, automation, evaluation, and support of the collaborative effort.

The guideline interchange format: a model for representing guidelines.

OBJECTIVE: To allow exchange of clinical practice guidelines among institutions and computer-based applications. DESIGN: The GuideLine Interchange Format (GLIF) specification consists of GLIF model and the GLIF syntax. The GLIF model is an object-oriented representation that consists of a set of classes for guideline entities, attributes for those classes, and data types for the attribute values. The GLIF syntax specifies the format of the test file that contains the encoding. METHODS: Researchers from the InterMed Collaboratory at Columbia University, Harvard University (Brigham and Women's Hospital and Massachusetts General Hospital), and Stanford University analyzed four existing guideline systems to derive a set of requirements for guideline representation. The GLIF specification is a consensus representation developed through a brainstorming process. Four clinical guidelines were encoded in GLIF to assess its expressivity and to study the variability that occurs when two people from different sites encode the same guideline. RESULTS: The encoders reported that GLIF was adequately expressive. A comparison of the encodings revealed substantial variability. CONCLUSION: GLIF was sufficient to model the guidelines for the four conditions that were examined. GLIF needs improvement in standard representation of medical concepts, criterion logic, temporal information, and uncertainty.

An empirical investigation into the conceptual structure of chest radiograph findings.

A method of investigating the conceptual structure of findings is presented, in which: 1) finding statements are extracted manually from free text, 2) the main concepts in each finding are manually identified and classified, and 3) the resulting sets of classes are examined for insights into finding structure. This study applies the method to chest radiograph reports. Although the subjects of the findings studied fall into seven classes, the same conceptual structure can be used for most of them.

The findings--diagnosis continuum: implications for image descriptions and clinical databases.

As part of the Unified Medical Language System (UMLS) project, we have been exploring the use of semantic net representation to build a medical ontology that can adapt to the needs and perspective of differing kinds of users with varying purposes. A principal objective is to facilitate indexing and retrieval of objects in a variety of target databases, using their own source vocabularies, while maintaining the representation of concepts to which these source vocabularies refer in a single consistent form, so that retrievals that span resource types can be accommodated. In addition, a particular area of deficiency of the existing UMLS Metathesaurus is that of clinical findings, a part of the problem being the multiple alternative views and granularity levels at which clinical findings are described in different target databases. The problem is particularly obvious when one examines the way in which image findings are described, which may be at a purely perceptual level, or at varying levels of aggregation into higher level observations or interpretations. We have developed a recursive model for representing observations and interpretations in a semantic net along a continuum of degree of aggregation, that appears to lend itself well to adaptation to varying perspectives.