EHR Data Similarity Scoring via Ontology Clustering
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing electronic health record systems face challenges in ensuring accuracy and completeness of recordkeeping, particularly due to the sheer volume of codes in medical ontologies like ICD-10-CM and ICD-10-PCS, which hinders searching and meaningful analysis, and care providers may struggle to recall relevant information due to the volume of patients and time elapsed since previous similar cases.
Innovation Solution
A method for extracting data from electronic health records to provide provider and patient data similarity scoring by encoding problem lists with concepts from a common ontology, parsing these concepts into clusters or categories, and calculating distances between providers or patients to identify similarities.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If the volume of codes in medical ontologies is increased to improve recordkeeping accuracy and completeness, then the precision of data encoding is improved, but the difficulty of searching and meaningful analysis increases
Solution Approach 1:
The patent segments the large ontology code space into smaller, manageable clusters or groups. Instead of presenting all 70,000+ ICD-10-CM diagnosis codes individually, the system organizes them into hierarchical structures (e.g., chapters, categories, clusters) that allow providers to navigate and search more efficiently while maintaining the precision of individual code encoding.
2Reliability
If the number of patients seen by a provider increases to improve experience and knowledge base, then the quality of care through experience is improved, but the provider's ability to recall relevant information deteriorates
Solution Approach 1:
The patent creates a computational model that copies and stores the provider's treatment patterns, decision-making processes, and care approaches for various patient conditions. This digital copy serves as an external memory system that can retrieve relevant information without relying on the provider's human recall, effectively scaling their experience across increasing patient volumes.
3Loss of information
If the volume of electronic health record data is increased to improve comprehensiveness of patient information, then the completeness of medical records is improved, but the time required to analyze and extract meaningful information increases
Solution Approach 1:
The patent performs preliminary actions by pre-processing and organizing patient data into structured formats with standardized ontologies during data entry and storage. Clinical decision support rules, treatment protocols, and analysis algorithms are pre-configured and loaded into the system, so that when analysis is needed, the system can quickly retrieve and process only the relevant pre-organized information rather than searching through raw unstructured data.
Data Source
AI summary
A system and method for extracting data from an electronic health record to provide provider and patient data similarity scoring includes: encoding a problem list for a plurality of patients with concepts from a common electronic health record ontology. In one aspect, the patients have electronic health records maintained by a plurality of providers. The system and method then may parse the concepts into a plurality of clusters or categories and determining, for each of the providers, a total number of patients that have at least one problem in a cluster or category or determining, for each patient, which of the plurality of clusters or categories correspond to at least one concept encoded in the patient's problem list. The system and method then may calculate for each pair of providers or patients, a distance between the providers or patients.