Biological Sensor Data Imputation for Intervention-Aware Causal Inference
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for imputing missing biological data fail to consider changes due to intervention measures and other influential factors, leading to a decrease in accuracy of causal inference.
Innovation Solution
An information processing system that includes a calculation device and storage device, which uses a causal relationship to accurately predict missing data by generating a trained model through a data imputation method considering confounder adjustments and intervention effects.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of information
If traditional imputation methods (filtering or machine learning models) are used to fill missing biological data, then the data completeness is improved, but the accuracy of causal inference decreases because changes due to intervention measures and influential factors are not considered
Solution Approach 1:
The patent changes the parameters considered in the imputation model by incorporating intervention measure data and confounder information into the machine learning model. This allows the model to account for influential factors that affect the missing data, thereby maintaining both data completeness and causal inference accuracy
Solution Approach 2:
The patent introduces an intermediary mechanism by using a trained machine learning model that specifically processes intervention measure data and confounder information. This intermediary model acts as a bridge to generate imputed values that reflect the underlying causal relationships, preventing the loss of causal inference accuracy
Data Source
AI summary
An information processing system includes: a processor that executes a program; and a storage device that stores the program, in which the storage device stores sensor data that is biological information of a user measured by a sensor and user data collected in association with the sensor data, and the processor acquires the sensor data, and imputes the sensor data using a causal relationship between the acquired sensor data and the user data.