AI Prediction Models with LLM Input Flexibility and SHAP Explanations
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
AI medical prediction models face limitations in handling arbitrary input data formats and lack detailed explanations for their output predictions, restricting their application and credibility in healthcare.
Innovation Solution
Integrate a large language model with a machine learning model to enable flexible input and output in natural language, using SHAP analysis to generate detailed explanations through Beeswarm, Partial Dependence, and Force plots.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If AI medical prediction models use specific data formats (JSON, etc.), then the model structure can be simplified, but the adaptability to handle arbitrary input formats is reduced
Solution Approach 1:
The patent introduces a large language model as an intermediary component between the user input and the traditional AI medical prediction model. This mediator translates arbitrary natural language inputs into the structured formats required by the prediction model, thereby maintaining the simplicity of the original model structure while significantly enhancing input format flexibility and adaptability.
2Device complexity
If AI medical prediction models output only probability values in specific formats, then the output structure remains simple, but the credibility and interpretability for users are reduced
Solution Approach 1:
The patent merges the traditional AI medical prediction model with a large language model to create an integrated system. This combination allows the system to maintain the probabilistic prediction capabilities of the original model while adding natural language explanation capabilities through the large language model, thereby enhancing prediction credibility and interpretability without significantly complicating the overall output structure.
3Reliability
If AI medical prediction models provide detailed explanation content, then the credibility and user understanding improve, but the computational complexity and processing time increase
Solution Approach 1:
The patent implements partial explanation generation by the large language model, focusing on providing key interpretive insights rather than exhaustive analyses. This partial action approach delivers sufficient credibility-enhancing explanations while avoiding the full computational overhead of generating comprehensive detailed analyses, thus balancing trustworthiness with processing complexity.
Data Source
AI summary
A method for establishing an artificial intelligence prediction model that integrates with a large language model (LLM) in the field of artificial intelligence is provided. The method for establishing the artificial intelligence prediction model includes the creation of the prediction model, obtaining predictions using the model, interpreting the results using SHAP analysis, and leveraging a large language model to overcome the format limitations encountered when dealing with input and output data. A system of an artificial intelligence prediction model is also provided.