EHR Search Recommendations Using ICD Code Co-Occurrence
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Clinicians face challenges in efficiently retrieving relevant information from Electronic Health Records (EHRs due to time constraints and cumbersome manual searching, necessitating improved methods for accurate search term recommendations.
Innovation Solution
A hybrid collaborative filtering model that generates search term recommendations based on co-occurrence frequencies between ICD codes and previous search terms, using matrix factorization to learn latent representations and calculate recommendation scores.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of information
If clinicians manually search through EHRs to retrieve patient information, then they can access detailed health records, but it consumes significant time and effort
Solution Approach 1:
The system pre-calculates and stores co-occurrence frequencies between ICD codes and search terms in advance. When a clinician needs information, the recommendation engine quickly retrieves pre-computed suggestions based on the current ICD code, eliminating the need for manual searching through patient records.
Solution Approach 2:
The patent replaces manual mechanical searching through EHR interfaces with an automated computational recommendation system. The system uses matrix factorization and co-occurrence analysis to automatically generate search term recommendations, substituting clinician manual effort with algorithmic processing.
2Ease of operation
If clinicians use built-in EHR search functions to find patient information, then they can retrieve specific data, but the process remains repetitive and cumbersome
Solution Approach 1:
The system enables the EHR to serve itself by automatically generating search term recommendations based on the current clinical context (ICD code). The system proactively identifies and suggests relevant search terms without requiring clinicians to manually formulate queries, making the system self-optimizing.
Solution Approach 2:
The system learns from historical search patterns and co-occurrence data to continuously improve recommendation quality. By analyzing which search terms are frequently associated with specific ICD codes, the system provides increasingly accurate suggestions that adapt to clinical needs over time.
3Reliability
If clinicians browse through patient records to find relevant information, then they can comprehensive review, but it is time-consuming under clinical time pressure
Solution Approach 1:
The system pre-computes co-occurrence statistics between ICD codes and search terms from historical data before clinical use. This preliminary analysis creates a knowledge base that enables rapid recommendation generation during patient encounters, ensuring both accuracy and speed.
Solution Approach 2:
The patent introduces an intermediary recommendation layer between the clinician and the EHR database. Instead of directly querying the database, the system first generates targeted search term suggestions based on the current ICD code, acting as a smart intermediary that filters and prioritizes potential information needs.
Data Source
AI summary
Disclosed embodiments include computer-implemented methods and systems that can efficiently generate highly relevant recommended search terms to clinicians. A hybrid collaborative filtering model recommends search terms for a specific patient to the clinician. The model draws on information from patients' clinical encounters and the searches that were performed during the clinical encounters. To generate recommendations, the model uses search terms which are (1) frequently co-occurring with the ICD codes recorded for the patient and (2) highly relevant to the most recent search terms. One variation of the model uses only the most recent ICD codes assigned to the patient. Another variation uses all ICD codes. Comprehensive experiments of embodiments of the methods and systems have demonstrate high levels of performance.