Medical Report Structuring via Search and Classification Algorithms
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Medical reports often contain complex, lengthy, and inconsistent descriptions of medical findings, making them difficult to read and understand, and may contain inaccuracies due to variations in reporting styles and adherence to standardized formats.
Innovation Solution
A computer-implemented method that receives a medical report and associated medical images, uses a searching algorithm to identify and structure medical findings, and applies a classification algorithm to verify and complement these findings, ensuring accuracy and consistency.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If speech recognition and free text are used for medical reporting, then the reporting process is simplified and faster, but the descriptions become unnecessarily complicated and long, making them difficult to read and understand
Solution Approach 1:
The patent segments the unstructured free text into structured components by identifying specific finding types (e.g., bone fractures, soft tissue injuries) and organizing them into standardized categories. This segmentation transforms the complicated continuous text into discrete, readable units that maintain the speed advantage of speech recognition while improving clarity.
Solution Approach 2:
The patent changes the parameter of text structure from unstructured free text to structured formatted output with standardized headings and categories. This parameter change converts the complicated narrative style into an organized format with clear sections, making the report easier to read while preserving the rapid documentation capability.
2Adaptability or versatility
If free text reporting is used, then the reporting process is more flexible, but inconsistencies and inaccuracies occur due to variations in reporting styles
Solution Approach 1:
The patent implements a universal structured template that can accommodate multiple types of medical findings (bone, soft tissue, organ injuries) within a single standardized format. This universal structure maintains flexibility in capturing diverse findings while ensuring consistency through standardized headings, classification systems, and reporting categories across all reports.
Solution Approach 2:
The patent incorporates verification mechanisms where the structured format provides feedback on completeness and accuracy. The standardized categories act as a checklist ensuring all relevant findings are captured, and the structured organization allows for systematic review, reducing inconsistencies while maintaining the adaptability to document various injury types.
3Loss of information
If detailed free text descriptions are provided, then comprehensive medical information is captured, but the reports become longer and more difficult to understand
Solution Approach 1:
The patent segments comprehensive medical information into organized sections with standardized headings (e.g., bone findings, soft tissue findings, organ findings). Each segment contains detailed information about specific finding types, maintaining completeness while improving understandability through clear visual organization and systematic categorization of the detailed content.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
A computer-implemented method for determining at least one structured medical finding in a medical report (1), the method comprising the following steps: (a) receiving the medical report (1), wherein the medical report (1) comprises text segments (6) describing medical findings, and receiving at least one medical image on which the medical report (1) is based; (b) applying a searching algorithm to the medical report (1) in order to identify medical findings in the medical report (1) and determine corresponding structured medical findings; (c) at least in case the determination of a structured medical finding by the searching algorithm is fulfils a pre-defined ambiguity criterion, applying a classification algorithm to the medical image to verify and/or complement the structured medical finding, wherein the output of the classification algorithm is at least one probable structured medical finding, and, if applicable, adjusting the structured medical finding determined by the searching algorithm on the basis of the output of the classification algorithm.