Graph-Based Clinical Concept Mapping for ICD Code Normalization
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The transition from ICD-9 to ICD-10 codes poses challenges due to the absence of clear one-to-one mappings, leading to orphan codes that hinder interoperability and require manual expert intervention for accurate clinical concept mapping, which is time-consuming and costly.
Innovation Solution
A graph-based clinical concept mapping algorithm that leverages SNOMED CT ontology to map ICD-9 and ICD-10 codes to unified SNOMED concepts, using a novel graph-based search algorithm and natural language processing to identify optimal mappings and group codes into higher-order concepts, thereby improving interoperability.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If manual expert intervention is used for clinical concept mapping between ICD-9 and ICD-10 codes, then mapping accuracy is improved, but time consumption and cost increase
Solution Approach 1:
The patent introduces SNOMED CT as an intermediary ontology that both ICD-9 and ICD-10 codes can map to. This mediator enables automated bidirectional mapping between ICD versions by providing a common reference framework, eliminating the need for direct manual mapping while maintaining clinical accuracy through the structured relationships in SNOMED CT.
Solution Approach 2:
The patent replaces the manual mechanical process of expert intervention with an automated computational system. The algorithm automatically traverses the SNOMED CT graph structure, computes mapping paths, and generates code mappings without human involvement, substituting the mechanical expert review process with algorithmic computation.
2Adaptability or versatility
If direct mapping between ICD-9 and ICD-10 codes is attempted, then interoperability is improved, but mapping completeness deteriorates due to orphan codes
Solution Approach 1:
SNOMED CT serves as a mediator that receives mappings from both ICD-9 and ICD-10 codebases. Orphan codes from either version can be mapped to SNOMED CT concepts, and then cross-referenced through the shared ontology, enabling complete mapping coverage even when direct ICD-to-ICD mappings don't exist.
Solution Approach 2:
The patent introduces a third dimension (SNOMED CT ontology) to the mapping problem. Instead of attempting direct one-dimensional mapping between ICD-9 and ICD-10, the system elevates the mapping to two dimensions by introducing SNOMED CT as an intermediate layer, allowing codes from either version to converge on shared clinical concepts.
3Measurement precision
If ICD codes are mapped to granular SNOMED concepts, then mapping precision is improved, but data aggregability and interpretability worsen
Solution Approach 1:
The patent enables dynamic aggregation of SNOMED CT concepts based on research needs. The system can traverse the hierarchical structure to aggregate granular mapped concepts into higher-level parent concepts when broader categorization is required, allowing the same mapping infrastructure to serve both precise and aggregated analysis requirements.
Solution Approach 2:
The SNOMED CT ontology provides multi-functionality by serving both as a granular mapping target for precise code alignment and as a hierarchical structure for aggregated data analysis. The same mapped dataset can be queried at different levels of granularity depending on the analytical needs, making the system universally applicable to both detailed and summary-level research questions.
Data Source
AI summary
A graph-based clinical concept mapping algorithm maps ICD-9 (International Classification of Disease, Revision 9) and ICD-10 (International Classification of Disease, Revision 10) codes to unified Systematized Nomenclature of Medicine (SNOMED) clinical concepts to normalize longitudinal healthcare data to thereby improve tracking and the use of such data for research and commercial purposes. The graph-based clinical concept mapping algorithm advantageously combines a novel graph-based search algorithm and natural language processing to map orphan ICD codes (those without equivalents across codebases) by finding optimally relevant shared SNOMED concepts. The graph-based clinical concept mapping algorithm is further advantageously utilized to group ICD-9/10 codes into higher order, more prevalent SNOMED concepts to support clinical interpretation.