Tongue Imager Model Update via Doctor Correction Feedback
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Tongue imagers in Traditional Chinese Medicine lack a mechanism for continuous updates and improvement, limiting their accuracy and adoption beyond initial manufacturing, as they rely on static software algorithms without a product update mechanism.
Innovation Solution
A method, client, and server system that continuously learns from doctor corrections to improve image analysis and diagnosis accuracy, using image analysis and assistant diagnosis models that can update based on new data and feedback, integrating machine learning for improved performance over time.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If static software algorithms are used in tongue imagers, then the device structure remains simple and manufacturing is easier, but the diagnostic accuracy and adaptability deteriorate over time without update mechanisms
Solution Approach 1:
The patent transforms the static software algorithm into a dynamic system that can continuously learn and update. The image analysis model and assistant diagnosis model are designed to accept new training data and iteratively improve their performance, making the diagnostic system adaptive rather than fixed.
Solution Approach 2:
The patent implements a feedback mechanism where correction information from doctors is collected and used to generate training data. This feedback loop allows the system to learn from actual diagnostic corrections and continuously improve its accuracy, addressing the limitation of static algorithms.
2Reliability
If continuous learning mechanisms are implemented, then diagnostic accuracy improves over time, but the device complexity and data processing requirements increase
Solution Approach 1:
The patent introduces an intermediary data processing layer that collects correction information, generates training data, and updates models. This intermediary mechanism manages the complexity by structuring the learning process into discrete, manageable steps rather than requiring complex real-time learning infrastructure.
Solution Approach 2:
The patent performs preliminary actions by collecting and organizing correction information into structured training data before model updates. This preparation work is done in advance, allowing the system to learn incrementally without requiring complex real-time processing during diagnosis.
3Adaptability or versatility
If correction information from doctors is integrated, then the system adaptability and diagnostic precision improve, but the operation complexity and data management burden increase
Solution Approach 1:
The patent implements self-service by automatically generating training data from correction information and updating models without manual intervention. The system autonomously manages the complexity of data integration and model training, reducing the operational burden on users while maintaining high adaptability.
Solution Approach 2:
The patent employs periodic action by updating models at scheduled intervals rather than in real-time. This approach allows the system to accumulate correction information and process updates periodically, maintaining operational simplicity while still achieving continuous improvement in adaptability.
4Measurement precision
If multiple models (image analysis and assistant diagnosis) are used, then measurement precision and diagnostic reliability improve, but the device complexity and computational requirements increase
Solution Approach 1:
The patent segments the diagnostic system into two distinct models: an image analysis model for processing tongue images and an assistant diagnosis model for generating diagnostic suggestions. This segmentation allows each model to specialize in specific tasks, improving precision while managing complexity through modular architecture.
Data Source
AI summary
The present disclosure relates to a method, a client, a server and a system for detecting a tongue image, and a tongue imager. The method for detecting the tongue image, which is applied to the client for detecting the tongue image, comprises: acquiring the tongue image, recognizing the tongue image by using an image analysis model, and generating and displaying a first label information; determining, after acquiring a correction information for the first label information, whether to adopt the correction information in accordance with an accuracy corresponding to the correction information, and generating a second label information; and analyzing the tongue image and the second label information by using an assistant diagnosis model, and generating an assistant diagnosis result.