Automated Clinical Summarization via NLP Screening and Template Population
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for generating clinical summarization reports are inefficient and prone to missing relevant data, relying on manual processes by nurses and yielding inferior results when using natural language processing algorithms.
Innovation Solution
A clinical summary management system that converts electronic medical records into a common format, applies a natural language processing algorithm to extract and screen data, and populates it into templates for automated generation of clinical summarization reports, enhancing accuracy and efficiency.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If manual clinical summarization by nurses is used, then patient information can be collected and synthesized, but the process is inefficient and time-consuming
Solution Approach 1:
The system enables automated clinical summarization where the computing system independently retrieves, converts, processes, and generates summary reports without requiring manual intervention by nurses, thereby eliminating the time-consuming manual process while maintaining summarization quality
Solution Approach 2:
The patent replaces the manual mechanical process of nurses reviewing and copying patient information with an automated natural language processing system that uses algorithms to extract, screen, and synthesize clinical data from electronic medical records
2Reliability
If manual clinical summarization is used, then relevant data can be identified, but relevant data is often missed negatively impacting patient care
Solution Approach 1:
The system employs a multi-stage feedback process where the NLP algorithm first extracts summarization data, then screens it based on multiple factors including relevance to diagnosis codes and disease types, and finally populates templates with validated information, ensuring comprehensive and accurate data identification
Solution Approach 2:
The natural language processing algorithm performs multiple functions simultaneously - retrieving data from various electronic medical record formats, converting them to a common format, extracting relevant information, screening based on multiple criteria, and populating standardized templates, ensuring no relevant data is missed
3Extent of automation
If natural language processing algorithms are applied to process complicated text, then automated analysis is achieved, but computer processing issues result in inaccurate understanding
Solution Approach 1:
The patent segments the text processing into distinct stages: first converting electronic medical records to a common format, then extracting summarization data, subsequently screening the extracted data based on multiple factors, and finally populating templates. This segmentation allows each stage to be optimized independently, improving overall accuracy
Solution Approach 2:
The system performs preliminary conversion of diverse electronic medical record formats into a common format before applying the NLP algorithm, and pre-identifies relevant factors for screening based on diagnosis codes and disease types, which prepares the data in advance and improves the accuracy of automated text understanding
Data Source
AI summary
Methods, non-transitory computer readable media, and devices that convert into a common electronic format a plurality electronic medical records retrieved in response to a request with identification data. A natural language processing algorithm is applied to obtain a subset of summarization data from each of the converted medical electronic record based on medical information data in the received request. The algorithm screens the initial subset of summarization data based on one or more factors to generate a reduced subset of summarization data for each of the converted medical electronic records. At least a portion of the reduced subset of summarization data is populated into data fields within one of a plurality of templates identified for each of the converted electronic medical records from the reduced subset of summarization data. A clinical summarization record is generated based on at least the populated summarization data in each of the identified ones of the plurality of templates. The clinical summarization record is provided in response to the received request.