AI Medical Record Screening for Coding Review Triage
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The process of reviewing health care records for medical coding is time-consuming and inefficient, often resulting in unnecessary labor when records lack supporting diagnoses, as coders spend hours searching for non-existent codes.
Innovation Solution
A decision support system utilizing two AI models trained differently: one based on medical term frequency and another on encounter-specific terms, generating combined recommendations to determine if a coding review is necessary, thereby reducing unnecessary reviews.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If coders manually review all medical records to ensure accurate coding, then coding accuracy is improved, but time consumption and labor costs increase significantly
Solution Approach 1:
The system performs preliminary analysis of medical records using AI models before human coders review them. The first AI model analyzes medical term frequency throughout the record, while the second AI model analyzes medical terms in encounter portions. This preliminary action identifies records likely to contain risk-adjusting diagnoses, so coders only need to review pre-screened records, reducing time consumption while maintaining accuracy.
Solution Approach 2:
The patent introduces an intermediary AI-based decision support system between the medical records and human coders. This intermediary automatically screens records and generates recommendations, acting as a mediator that filters out records unlikely to contain relevant diagnoses. This reduces the burden on human coders while preserving coding accuracy for records that actually need review.
2Reliability
If coders review lengthy medical records thoroughly to identify supporting diagnoses, then coding completeness is improved, but productivity decreases
Solution Approach 1:
The system performs preliminary screening of medical records using two AI models that analyze medical term frequency and encounter-specific terms. This preliminary action identifies records with high probability of containing risk-adjusting diagnoses, allowing coders to focus their thorough review efforts only on these pre-identified records rather than all records, thus improving productivity while maintaining coding completeness.
Solution Approach 2:
The system applies partial action by not requiring complete manual review of all medical records. Instead, it uses AI models to perform partial analysis (term frequency analysis and encounter portion analysis) to identify promising records, then applies full human review only to those specific records. This selective approach improves overall productivity while ensuring coding completeness for records that actually contain relevant diagnoses.
3Measurement precision
If multiple AI models are used to analyze medical records, then identification accuracy is improved, but system complexity increases
Solution Approach 1:
The system segments the analysis task into two distinct AI models with different functions: the first AI model analyzes medical term frequency throughout the entire record, while the second AI model analyzes medical terms specifically in encounter portions. This segmentation allows each model to specialize in a particular aspect of record analysis, improving overall identification accuracy while keeping each individual model relatively simple and manageable.
Data Source
AI summary
A method includes receiving a record containing clinical information associated with a patient; processing the record using a first AI model that is trained based on medical term frequency throughout respective ones of a plurality of historical medical records; processing the record using a second AI model that is trained based on medical terms used in encounter portions of the respective ones of the plurality of historical medical records; generating a first recommendation for performing a coding review of the record based on processing the record using the first AI model; generating a second recommendation for performing the coding review of the record based on processing the record using the second AI model; and logically combining the first and second recommendation for performing the coding review to generate a final recommendation for performing the coding review.