AI Medical Profile Classifier for Patient Risk Identification
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
There is a need for new techniques to effectively analyze medical information for a population of subjects, as existing methods have limited progress in utilizing the wealth of data available for patient medical histories.
Innovation Solution
A method and system utilizing an artificial intelligence system to identify subjects potentially impacted by a medical condition. This involves defining experimental and control groups based on specific criteria related to the medical condition, training the AI system with these criteria and secondary characteristics, and using the trained classifier to mark subjects in the database as potentially affected by the medical condition.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of information
If medical information is stored on computers and communicated over networks, then availability and accuracy of medical information improve, but security risks and information theft concerns increase
Solution Approach 1:
The patent introduces an intermediary AI system that processes medical information locally without requiring direct network communication for analysis. The system uses local computing resources to perform machine learning inference, acting as a mediator between stored medical data and diagnostic outcomes, thereby reducing security risks associated with network transmission while maintaining information availability
Solution Approach 2:
The patent creates local copies of medical information and analysis models that can be processed without constant network access. By storing medical profiles locally and using local AI models, the system eliminates the need to transmit sensitive data over networks, thus maintaining availability while reducing security vulnerabilities
2Object-affected harmful factors
If traditional paper-based medical information storage is used, then security concerns are reduced, but accessibility and analysis capability are limited
Solution Approach 1:
The patent replaces the mechanical paper-based storage system with a computerized database system that uses structured digital profiles. This substitution enables advanced querying, searching, and analysis capabilities while implementing security measures at the software and data architecture level, thus improving accessibility without compromising security
3Loss of information
If computerized analysis of medical information is implemented, then accuracy and availability improve, but the complexity of analysis techniques required increases
Solution Approach 1:
The patent transforms unstructured or semi-structured medical information into standardized structured profiles with defined parameters and data types. This parameterization enables the use of simpler, more efficient machine learning algorithms and database queries, reducing the complexity of analysis techniques while maintaining or improving accuracy through consistent data formatting and validation
Data Source
AI summary
Subjects who are potentially impacted by a medical condition are identified. An experimental group includes subjects having a positive indication for a specific criterion related to the medical condition in their medical profiles. A control groups includes subjects having a negative indication for the specific criterion. An artificial intelligence system is trained using the specific criterion and secondary characteristics of the subjects of the experimental and control groups to construct a classifier for the medical condition. The classifier is used to extract a target group of subjects from a population of subjects. A medical profile of each subject of the target group is marked as potentially affected by the medical condition. A system includes the artificial intelligence system and a database for storing the medical profiles. Deep learning or machine learning may be used to analyze medical images such as retinal images.