Dental Chart Generation From Medical Records Using LLM Intermediates
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing medical record systems fail to consider patient comprehension, leading to cognitive differences between physicians and patients, as patients rely solely on physician explanations for understanding their conditions.
Innovation Solution
A method and system utilizing large language models to generate a teeth position chart from medical records, incorporating image generation to visually represent dental conditions, enabling patients to directly comprehend their conditions without physician explanation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If medical record data is recorded using familiar terminology for physician convenience, then recording efficiency is improved, but patient comprehension deteriorates due to cognitive differences
Solution Approach 1:
The system generates a visual copy (teeth position chart) of the medical record data that patients can directly comprehend. The image generating module creates an visual representation copying the essential information from the structured data, allowing patients to understand their conditions without relying on physician explanations while preserving the efficient terminology-based recording method.
Solution Approach 2:
The system transforms the one-dimensional text-based medical record data into a two-dimensional visual chart. By converting structured data about tooth positions and conditions into a spatial visual representation, the system enables patient comprehension without changing the efficient recording methodology, resolving the contradiction between recording efficiency and patient understanding.
2Loss of information
If visual charts are generated to improve patient comprehension, then information accessibility is improved, but system complexity increases due to additional processing modules
Solution Approach 1:
The system introduces an intermediate file format as a mediator between the text-based medical record data and the visual chart generation. The image generating module first converts structured data into this intermediate format, which then serves as input for visual chart creation. This intermediary step simplifies the overall process by creating a standardized bridge between data representation and visual visualization.
3Measurement precision
If historical data comparison is performed to improve accuracy, then diagnostic precision is improved, but processing time increases
Solution Approach 1:
The system performs preliminary actions by pre-storing medical record historical data and result historical data in separate databases. When generating a teeth position chart, the system quickly retrieves and compares against this pre-organized historical data rather than performing complex analysis from scratch. This preliminary organization of data enables fast comparison while maintaining high diagnostic precision.
Solution Approach 2:
The system replaces manual physician comparison with automated computer-based comparison algorithms. The processing module automatically compares the current medical record data against historical data in databases, using computational methods to quickly identify patterns and discrepancies. This substitution of automated processing for manual analysis maintains precision while significantly reducing processing time.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
The present application provides a method for generating a dental chart. The method includes: inputting a medical record data into a first large language model; analyzing the medical record data by the first large language model and then generating and outputting an intermediate file corresponding to the medical record data; inputting the intermediate file into a image generating model; and analyzing the intermediate file by the image generating model and then generating and outputting the dental chart corresponding to the intermediate file. The intermediate file has at least one dental position and at least one symptom information corresponding to the at least one dental position. Thereby, the method can convert a patient's medical record data into a corresponding dental chart, in order to directly know the patient's condition. In addition, a computing device and a non-transitory computer-readable recording medium for generating the dental chart are also provided.