Data processing: artificial intelligence – Knowledge processing system
Reexamination Certificate
1999-10-29
2002-08-20
Davis, George B. (Department: 2121)
Data processing: artificial intelligence
Knowledge processing system
C706S046000, C600S300000, C705S003000
Reexamination Certificate
active
06438533
ABSTRACT:
BACKGROUND OF THE INVENTION
The present invention generally relates to the compilation, storage, transfer, retrieval, and aggregation of information. In particular, the invention is an apparatus designed to compile, store, transfer, retrieve, and/or aggregate electronic medically-related records.
The medical community has long sought a method for coordinating and analyzing aggregations of medically-related records. Technical difficulties have prevented the community from designing and implementing a system that permits computerized analysis of medical records. Because medical records are generally non-numerical expressions of concepts and because there is no uniform method of expressing those concepts, computers have traditionally lacked the capacity to recognize when a given medical record satisfies a query.
The community has attempted to design data structures that facilitate computerized classification of medical concepts. A proper data structure must balance the objectives of permitting concepts to be designed with a degree of expressivity, and of facilitating classification of concepts by limiting their complexity. Data structures in the prior art, however, have failed to provide the proper balance of expressivity and complexity to permit efficient coordination and retrieval of medical records.
Many data structures in the prior art lack adequate expressivity. They classify medical concepts according to an alphanumeric code organized in strict hierarchies. A strict hierarchy requires that a single concept have one meaning, one code number, and thus, only one parent concept. The concept must be positioned in one branch of a code-based hierarchical tree.
The SNOMED(™) International work of medical nomenclature is one example of a medical terminology structured according to a strict hierarchy. SNOMED International, which is incorporated herein by reference, is a systemized nomenclature of human and veterinary medicine, which is published, copyrighted and maintained by the College of American Pathologists. SNOMED International is an advanced nomenclature and classification of medical terms and codes.
SNOMED International, and its predecessor SNOMED, provide a detailed and specific coded vocabulary of names and descriptions used in healthcare. Their purpose is to index, store and retrieve information about a patient in a computerized medical record. The original SNOMED, published in 1974, consisted of six nomenclature categories or classes representing various aspects of the human being from a pathophysiologic point of view. These classes include, for example, topography (anatomy), morphology (descriptions of changes in the normal anatomy of the body), etiology (cause or causal agents of diseases or injuries, including drugs and chemicals), function (functions of the human body), disease/diagnosis (general and complex disease terms and syndromes), and procedures (administrative, preventive, diagnostic and therapeutic actions taken to prevent or cure).
SNOMED International, published in 1993, expanded the nomenclature to 11 categories and provided some general linkage modifiers. SNOMED and SNOMED International have been licensed to numerous computer software vendors that have developed customized database applications. An example of the hierarchical constraints of SNOMED International are shown in
FIG. 8
, which illustrates the classification of the category of diseases by organ system or by the underlying etiology. The present invention, as described in more detail below, eliminates the strictly hierarchical constraints of SNOMED International.
FIG. 9
illustrates one type of modification of the relationships of SNOMED International by the present invention.
Strict hierarchies based on alphanumeric codes provide less than satisfactory data structures for compiling, storing, transferring, retrieving, and/or aggregating medical records for two primary reasons. First, the data structure limits the efficacy of queries based upon hierarchical relationships, because medical concepts are complex ideas which are a function of the context of use, and therefore cannot be fully represented solely by a single code. Second, the data structure forecloses the user from forming queries based upon non-hierarchical relationships.
Strict hierarchical structures limit the ability of computers to search medical records according to hierarchical relationships because many medical concepts cannot be properly classified in a single position in a hierarchy. A term that can be described as belonging to several different groups must be encoded into one, and only one, group. One example is the term “pneumothorax. ” Pneumothorax can be described as a member of the group of terms “diseases of the respiratory system,” or a member of the group of terms “diseases of the pleura, mediastinum and diaphragm.” In SNOMED International, for example, the term “traumatic pneumothorax” is classified in the class of “injuries and poisonings,” with no link to the class of “diseases of the respiratory system.”
A computer cannot satisfactorily search, sort and retrieve medical records that are organized according to a strict hierarchy. A query for a collection of records classified under a given parent term will not retrieve all pertinent records, when those records contain concepts that were constrained to be positioned in a separate branch of the hierarchy. In the above example, a query in SNOMED International for records of patients with diseases of the lung will not retrieve records of patients afflicted with traumatic pneumothorax, because traumatic pneumothorax was classified under “injuries and poisonings.”
The prior art suffers from an additional limitation in that its data structures do not adequately permit users to search databases according to non-hierarchical queries. Strict hierarchies bundle the meaning of medical concepts into a single alphanumeric code. The code number expresses the relationship of the medical concept to other concepts along the same hierarchical axis, but fails to express the non-hierarchical characteristics of that concept. The code representing the medical concept does not provide information, for example, about how the concept manifests itself or how it is caused. Users cannot enter a query based upon non-hierarchical relationships when the meaning of a concept is contained only in a hierarchical code.
Some disclosed prior art database structures provide a limited ability to search medical records based upon non-hierarchical relationships. These systems, however, leave the meaning of the concepts based in an alphanumeric code, with non-hierarchical linking terms appended to the core meaning. These linking terms are not integrated into the meaning of the term, and do not explicitly define how the linking term relates to the core concept. For example, SNOMED International allows the concept “acute appendicitis” to be represented as an appendicitis that is acute, by allowing the user to select the alphanumeric code for the concept “appendicitis,” with the alphanumeric code for the concept “acute.” It remains unclear, however, whether the term “acute” refers to an appendicitis with an acute onset, or an appendicitis with an acute severity.
An ideal data structure for retrieval of medical records needs to offer limited complexity to properly process searches. A surfeit of linking terms restricts the ability of a computer to accurately process a query. The present invention addresses these concerns with a hierarchical structure that includes linkages between more specific child terms and less specific parent terms. At the same time, the present invention provides linkages along non-hierarchical axes or based on non-hierarchical relationships.
SUMMARY OF THE INVENTION
The foregoing limitations of prior art data retrieval systems are addressed by the following system for data retrieval of medical records.
The apparatus employs a terminology knowledge base with a data structure specifically designed to permit a classifier to execute subsumption checks upon a database of patient records. In on
Campbell Keith E.
Spackman Kent A.
Abrams Hugh A.
College of American Pathologists
Davis George B.
Sidley Austin Brown & Wood
LandOfFree
System for retrieval of information from data structure of... does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with System for retrieval of information from data structure of..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and System for retrieval of information from data structure of... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-2933768