String Normalization for Medical Ontology Classification
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Traditional systems are inadequate for classifying strings of words into complex ontologies or nomenclatures, such as SNOMED CT, due to their inability to handle variability in medical diagnoses expressed differently by various professionals, leading to ineffective classification of medical data.
Innovation Solution
A system and method for adaptive categorization of words that normalizes input strings by tagging, comparing them to a predetermined categorization scheme, and using baseform and feature transformations to identify matches, allowing for classification within complex hierarchically-organized nomenclatures.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional classification systems are used, then the system structure is simple, but the classification accuracy and adaptability to complex ontologies deteriorate
Solution Approach 1:
The patent segments the classification process into multiple distinct modules: normalization module, tagging module, literal matching module, baseform matching module, feature matching module, and feedback module. Each module handles a specific aspect of the classification task, allowing the system to achieve high classification accuracy for complex ontologies while maintaining manageable system complexity through modular design.
Solution Approach 2:
The system dynamically adapts its classification approach by attempting literal matching first, then progressively trying baseform matching and feature matching if literal matching fails. The system also incorporates feedback mechanisms that allow it to learn from mismatches and improve future classifications, enabling it to handle the variability in medical diagnosis expressions effectively.
2Ease of operation
If traditional broad categories are used for classification, then the system is easier to operate, but the classification effectiveness for complex ontologies deteriorates
Solution Approach 1:
The system performs preliminary normalization of the input string before classification, converting it to a standard format and identifying key features in advance. This preliminary processing simplifies the subsequent matching operations and improves classification effectiveness without requiring complex real-time processing during the actual classification step.
Solution Approach 2:
The patent introduces intermediate representations including tagged strings, baseforms, and feature vectors that serve as mediators between the raw input text and the final classification. These intermediaries simplify the matching process by transforming complex text variations into standardized forms that can be efficiently compared against the ontology, maintaining ease of operation while improving classification precision.
3Speed
If literal matching is used, then the processing speed is fast, but the adaptability to text variability deteriorates
Solution Approach 1:
The system dynamically selects the matching approach based on the input characteristics. It starts with fast literal matching for exact matches, then transitions to baseform matching and feature matching when text variability is detected. This dynamic adaptation allows the system to maintain high processing speed for common cases while achieving high adaptability for varied text expressions.
Solution Approach 2:
The system applies different matching strategies to different parts of the classification process: literal matching for exact phrase matches, baseform matching for morphological variations, and feature matching for semantic equivalence. This localized application of different matching qualities allows the system to optimize both speed and adaptability for different types of text variability.
4Measurement precision
If multiple matching stages are implemented, then the classification accuracy improves, but the processing time increases
Solution Approach 1:
The system performs preliminary normalization and feature extraction before the matching stages, so that when multiple matching stages are executed, they operate on pre-processed data. This preliminary action reduces the computational burden of subsequent matching operations, allowing the system to maintain high classification accuracy through multiple stages while minimizing the additional processing time required.
Solution Approach 2:
The system dynamically determines when to proceed to subsequent matching stages based on the results of previous stages. If literal matching succeeds, the system terminates early without executing baseform or feature matching. This dynamic control ensures that multiple matching stages only execute when necessary, maintaining high classification accuracy while avoiding unnecessary processing time expenditure.
Data Source
AI summary
The present invention relates generally to a system and method for categorization of strings of words. More specifically, the present invention relates to a system and method for normalizing a string of words for use in a system for categorization of words in a predetermined categorization scheme. A method for adaptive categorization of words in a predetermined categorization scheme may include receiving a string of text, tagging the string of text, and normalizing the string of text. Normalization may be performed with a three-stage algorithm including a literal match processing stage, an approximation match processing stage, and a nearest neighbor match processing stage. The normalized string of text can be compared to a number of sequences of text in the predetermined categorization scheme.