Patient Condition Prediction Model Using Synthetic Longitudinal Data
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing patient condition prediction models face challenges in improving prediction accuracy due to insufficient learning data, as collecting actual patient conditions over time is difficult, limiting the generation of sufficient learning data needed for model updates.
Innovation Solution
A model generation apparatus and method that obtain learning data including index values indicating patient conditions at different times, using a prediction model to forecast future conditions based on past conditions and change information, allowing for model updates that incorporate predicted future condition values.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If actual patient condition data is collected to generate learning data, then prediction accuracy can be improved, but the amount of available learning data is insufficient
Solution Approach 1:
The system performs preliminary actions by generating synthetic learning data through simulation before actual patient data collection. The simulation generates predicted future condition values and change information that are used to create learning datasets, enabling model training to proceed even when actual longitudinal patient data is insufficient.
Solution Approach 2:
The system creates copies of patient condition data by simulating future conditions based on current state and change tendencies. Instead of requiring actual repeated measurements of the same patient over time, the system generates synthetic copies of what the condition values would be, effectively multiplying the available learning data without additional clinical measurements.
2Measurement precision
If more learning data is collected to improve prediction accuracy, then model performance improves, but the complexity of data collection and measurement increases
Solution Approach 1:
The system performs self-service by using the prediction model itself to generate the learning data needed for its own training and improvement. The model generates predicted future condition values and change information that feed back into creating learning datasets, eliminating the need for external complex data collection systems to provide synthetic training data.
Solution Approach 2:
The system changes parameters by transforming current condition values and change tendencies into predicted future condition values through the prediction model. This parameter transformation generates synthetic learning data with different temporal characteristics without requiring physical changes to measurement systems or data collection procedures.
3Measurement precision
If prediction accuracy is improved by using more actual patient data, then the model becomes more reliable, but the time required to collect sufficient data increases
Solution Approach 1:
The system performs preliminary data generation actions by simulating patient condition trajectories in advance. Instead of waiting to collect actual patient data over extended periods, the system pre-generates learning datasets with predicted future conditions that can immediately be used for model training and validation.
Solution Approach 2:
The system adds a temporal dimension to data generation by creating synthetic future condition values based on change tendencies. This transforms static or limited cross-sectional patient data into pseudo-longitudinal learning datasets, effectively creating time-series learning data without requiring actual longitudinal observation periods.
Data Source
AI summary
A model generation apparatus includes: an acquisition unit for obtaining learning data including a first index value indicating a condition of a sample patient at a first time, and a second index value indicating a condition of the sample patient at a second time that is after the first time; and a learning unit for learning a prediction mode for predicting a condition of a target patient at the second time on the basis of a condition of the target patient at the first time, wherein the learning unit is configured to learn the prediction model by updating the prediction model on the basis of a third index value indicating a condition of the sample patient at the second time that is predicted by the prediction model on the basis of the first index value, and a change information indicating a change tendency of the condition over time.