AI Models Predict Genetic Mutations from Tumor Images
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for identifying the right patients for specific therapies in clinical trials and clinical settings are inefficient due to invasive and costly molecular testing, which can be invasive and resource-intensive, and often fail to account for drug resistance and non-responder patients.
Innovation Solution
The use of AI/ML models that leverage public datasets for clinical, genomics, histopathological, and radiomics data to develop algorithms that predict genetic mutations and molecular alterations, reducing the need for extensive testing by identifying relevant genetic mutations and excluding patients with resistance to specific therapies through systems biology modeling.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If molecular testing is performed to identify genetic mutations, then accuracy in identifying right patients for therapy is improved, but cost and invasiveness increase
Solution Approach 1:
The patent creates a virtual copy of the molecular testing process through AI/ML models that simulate and predict genetic mutations based on clinical and imaging data, replacing the need for physical tissue sampling and laboratory testing while maintaining diagnostic accuracy
Solution Approach 2:
The patent introduces AI/ML models as an intermediary between clinical/imaging data and genetic mutation identification, using these models to predict mutations without direct molecular testing, thereby eliminating the need for invasive tissue acquisition
2Reliability
If extensive molecular testing is performed to identify all genetic mutations, then completeness of genetic profile is improved, but resource consumption and time increase
Solution Approach 1:
The patent applies partial action by using AI/ML models to predict only the most clinically relevant genetic mutations based on clinical and imaging data, rather than performing exhaustive testing for all possible mutations, thereby achieving sufficient completeness with reduced resource consumption
Solution Approach 2:
The patent performs preliminary prediction of genetic mutations using AI/ML models before committing to extensive molecular testing, allowing clinicians to prioritize or skip certain tests based on predicted results, thereby improving overall efficiency while maintaining completeness
3Reliability
If multiple molecular alterations are tested to account for drug resistance, then reliability of therapy selection is improved, but device complexity and cost increase
Solution Approach 1:
The patent merges multiple molecular alteration assessments into a single integrated AI/ML modeling framework that simultaneously evaluates various genetic mutations and predicts drug resistance, replacing multiple separate testing procedures with one unified system
Solution Approach 2:
The patent creates a universal AI/ML platform that can predict multiple different genetic mutations and resistance patterns across various cancer types using the same clinical and imaging data inputs, eliminating the need for separate testing systems for different alterations
Data Source
AI summary
Disclosed herein are methods and systems for identfying genetic mutation and molecular alterations via imaging and clinical proxies using machine learning techniques. A processor can receive an image of a tumor of a patient. The processor can execute a first model to identify one or more visual attributes of the tumor using the image of the tumor as input. The processor can execute a second model to predict a genetic mutation or molecular alterations of the patient using the one or more visual attributes as input. The processor can identify a therapy protocol associated with the tumor based on the genetic mutation.