Machine Learning Model Retraining via User Feedback
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Machine learning models struggle with generalization to unseen data, lack of user validation for proper behavior, and difficulty in providing feedback to improve performance, particularly in predicting measures of interest for coronary artery disease diagnosis.
Innovation Solution
The system allows for online retraining of machine learning models using user feedback on predicted measures of interest for secondary tasks, which improves the model's performance in predicting both primary and secondary tasks by incorporating user corrections and confidence measures.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If machine learning models are trained to predict measures of interest for coronary artery disease, then prediction capability is improved, but generalization to unseen data deteriorates
Solution Approach 1:
The system performs preliminary actions by training the machine learning model on multiple related tasks (primary task of predicting hemodynamic indices and secondary tasks of predicting anatomical features) before deployment. This multi-task preliminary training enables the model to learn transferable representations that improve generalization to unseen data while maintaining prediction accuracy for the primary clinical task.
2Measurement precision
If machine learning models predict hemodynamic indices directly, then primary diagnostic capability is improved, but user validation deteriorates
Solution Approach 1:
The system introduces secondary tasks (predicting anatomical features such as stenosis location, vessel segmentation, and measurement points) as intermediaries between the input medical images and the primary hemodynamic index predictions. These intermediate anatomical predictions serve as interpretable mediators that clinicians can visually validate against the images, thereby enabling user validation while maintaining accurate hemodynamic index prediction through the multi-task learning framework.
3Measurement precision
If machine learning models are trained for specific tasks, then task-specific performance is improved, but feedback integration deteriorates
Solution Approach 1:
The system implements feedback mechanisms where user corrections of the secondary task predictions (anatomical features) are fed back into the training process. This feedback loop allows the model to learn from user expertise and continuously improve its predictions for both secondary tasks and the primary hemodynamic index task, thereby integrating user feedback while maintaining task-specific performance through iterative retraining.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
Systems and methods for retraining a trained machine learning model are provided. One or more input medical images are received. Measures of interest for a primary task and a secondary task are predicted from the one or more input medical images using a trained machine learning model. The predicted measures of interest for the primary task and the secondary task are output. User feedback on the predicted measure of interest for the secondary task is received. The trained machine learning model is retrained for predicting the measures of interest for the primary task and the secondary task based on the user feedback on the output for the secondary task.