Automated SOAP Note Generation from Doctor-Patient Conversations
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Physicians spend a significant amount of time on EHR documentation, which leads to burnout, as manual entry of patient conversations is tedious and time-consuming, especially when dealing with lengthy conversations that are not efficiently processed by existing systems.
Innovation Solution
A data processing system that filters and pre-selects noteworthy utterances from patient-physician conversations to reduce the length of data input for machine learning models, significantly improving processing time and predictive performance by generating automated SOAP notes.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If the entire conversation transcript is fed into the machine learning model, then the model has access to complete information for accurate diagnosis prediction, but the processing time increases significantly due to the quadratic time complexity of the model
Solution Approach 1:
The patent extracts only the noteworthy utterances from the full conversation transcript that are most relevant to diagnosis and RoS abnormality prediction. This extraction process removes irrelevant portions of the conversation, reducing the input size to the machine learning model while retaining the essential information needed for accurate predictions, thereby resolving the contradiction between complete information access and processing efficiency
Solution Approach 2:
The patent segments the conversation transcript into individual utterances and identifies noteworthy ones based on specific criteria. This segmentation allows the system to process only relevant segments rather than the entire transcript, reducing computational load while maintaining prediction accuracy by focusing on diagnostically significant portions of the conversation
2Reliability
If manual EHR documentation is performed by physicians, then the documentation can be completed with full clinical judgment and accuracy, but the time required increases significantly leading to physician burnout
Solution Approach 1:
The patent implements an automated system that performs EHR documentation tasks without requiring physician intervention. The machine learning model automatically generates diagnosis predictions and RoS abnormality assessments from conversation transcripts, allowing the system to serve itself in completing documentation tasks that would otherwise require significant physician time and effort
Solution Approach 2:
The patent replaces the manual mechanical process of physician documentation with an automated computational system. The machine learning model substitutes for the physician's manual writing and cognitive processing, using algorithms to analyze conversation transcripts and generate structured documentation, thereby reducing the time burden on physicians while maintaining documentation quality
Data Source
AI summary
A data processing system accesses a digital resource that includes a plurality of sections and a classifier configured to detect contents representing one or more portions of a communication with increased likelihood of being cited as evidence associated with a particular one of the sections. The data processing system receives a stream of data items representing a communication and generates content for at least one of the sections. The data processing system parses one or more fields in the data items, extracts values from the one or more parsed fields, identifies, by the classifier, that the extracted values are represented in one or more portions of the contents representing the one or more portions of the communication with increased likelihood of being cited as evidence, identifies that the extracted values are associated with a particular section of the digital resource, and generates content for that particular section.