Counterfactual prediction model data processing method, device, equipment and storage medium

Abstract

The application discloses a counterfactual prediction model data processing method, device, equipment and storage medium, and relates to the technical field of artificial intelligence. The method includes: obtaining historical observation data related to the target event; constructing training samples based on the historical observation data; obtaining countermeasure prediction data and balance representation parameters through the encoding network based on the training samples; and obtaining countermeasure prediction data and balance representation parameters through the decoding network based on the balance representation parameters and countermeasures The real data is used to obtain the predicted data of the output variable; the training loss is determined based on the predicted data of the countermeasures, the real data of the countermeasures, the predicted data of the output variable and the label data; the model is trained based on the training loss. The training method provided by this application can eliminate the correlation between time-varying confounding factors and countermeasures, so that the deviation caused by time-varying confounding factors can be eliminated during counterfactual prediction, and the prediction accuracy of the model can be improved. The technical solution provided by this application is applicable to the counterfactual prediction of the epidemic.

Description

technical field [0001] The embodiments of the present application relate to the technical field of artificial intelligence, and in particular to a counterfactual prediction model data processing method, device, device, and storage medium. Background technique [0002] With the research and progress of artificial intelligence technology in counterfactual prediction models, researchers can use counterfactual prediction models to predict the effect of target response measures on epidemic control. [0003] Taking a counterfactual prediction model composed of a recurrent neural network as an example, train the counterfactual prediction model based on historical epidemic data to obtain a trained counterfactual prediction model, and then predict the future development trend of the epidemic by changing the input data of the model . For example, the response measures in the input data of the model are replaced with the target response measures to predict the control effect of the ta...

AI Technical Summary

Problems solved by technology

[0004] However, counterfactual prediction models based on target responses are biased and inaccurate due to the association between time-varying confounding factors and responses

Images