Adaptive Procedure Support System for Medical Workflows
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for identifying workflow steps in minimally invasive procedures can only recognize predefined categories, failing to account for user-specific or adaptive workflow steps, which limits personalized support for medical operators.
Innovation Solution
A computer-implemented method that captures time-resolved procedure data, applies a trained function using machine learning to identify target procedure configurations and provide adaptive support, including user-specific adjustments based on comparison of training and actual procedure configurations.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If fixed categories are used for workflow step identification, then the system operation becomes simpler and more standardized, but the ability to identify user-specific workflow steps is lost
Solution Approach 1:
The patent implements dynamic adaptability by enabling the system to adjust workflow categories and parameters based on individual user behaviors and preferences. The system transitions from static fixed categories to dynamic user-specific categories through continuous learning from user interactions, allowing workflow identification to adapt to each user's unique patterns while maintaining operational simplicity.
Solution Approach 2:
The patent changes the parameters of workflow identification from fixed predefined categories to flexible user-specific categories. By modifying the categorization parameters based on user behavior data and preferences, the system can identify workflow steps that are specific to each user while maintaining the overall system operation simplicity through automated parameter adjustment.
2Stability of the object's composition
If standardized workflow categories are implemented, then procedural consistency is improved, but personalized procedure support for individual operators is limited
Solution Approach 1:
The patent segments workflow categories into two levels: standardized common categories that ensure procedural consistency, and user-specific customized categories that provide personalized support. This segmentation allows the system to maintain stable standardized categories for general procedures while creating adaptive user-specific categories for individual operator preferences and patterns.
Solution Approach 2:
The patent introduces dynamic adaptation to standardized workflows by allowing the system to learn and adapt to individual user behaviors. The workflow categories dynamically adjust to incorporate user-specific patterns while maintaining the core standardized structure, enabling personalized procedure support without sacrificing overall procedural consistency.
3Productivity
If machine learning models are trained on general procedure data, then broad procedure recognition is achieved, but user-specific workflow patterns cannot be identified
Solution Approach 1:
The patent applies preliminary training with general procedure data to establish a foundation of broad procedure recognition capabilities. This preliminary training ensures the system can recognize common workflow patterns across different procedures, which then serves as a base for subsequent user-specific adaptation through continued learning from individual user interactions.
Solution Approach 2:
The patent implements feedback mechanisms where user-specific workflow patterns are continuously learned and incorporated into the machine learning model. The system receives feedback from user interactions and adjusts its recognition patterns accordingly, enabling it to identify user-specific workflows while maintaining the broad procedure recognition capability established through general training.
Data Source
AI summary
A method includes capturing time-resolved procedure data relating to a medical procedure, which includes procedure parameters and/or a device configuration and/or physiological data of an examination object and/or medical image data of the examination object. Time information relating to a target instant and a target procedure configuration for the target instant are provided by applying a trained function to input data. The input data is based on the procedure data. The target procedure configuration, including a target procedure parameter and/or a target device configuration, and the time information, designating an interval between an identified procedure event and the target instant and/or the target instant, is provided as the output data. At least one parameter of the trained function is adjusted based on a comparison of a training procedure configuration with a comparison procedure configuration and a comparison of training time information with comparison time information.