This research presents a novel approach that utilizes Deep Neural Network(DNN)technology to estimate glucose levels in enzyme-catalyzed oxidation reactions.Typically,glucose content is measured using glucose oxidase i...This research presents a novel approach that utilizes Deep Neural Network(DNN)technology to estimate glucose levels in enzyme-catalyzed oxidation reactions.Typically,glucose content is measured using glucose oxidase in biosensors.However,when this enzyme is used as a catalyst during the reaction,it may interfere with accuracy of the detection.To address this challenge,we propose a DNN-based method that incorporates an attention mechanism to improve glucose prediction accuracy.The study investigates the synthesis of zinc gluconate through enzyme-catalyzed processes,employing numerical simulations to analyze heat release and transfer in the reactor.By integrating the attention mechanism,the research enhances the predictive performance for glucose levels in these reactions.DNN models are known for their ability to learn complex patterns in data,and their capacity to handle large datasets makes them particularly suitable for predicting glucose levels in biochemical processes.This capability enables the model to effectively capture subtle relationships between variables.Experimental results demonstrate a significant improvement in predictive accuracy,with a mean absolute error(MAE)as low as 0.024,which is considerably better than conventional methods.The study also uses numerical simulations to analyze heat release and transfer during the reaction process and validates the model’s generalization ability through K-fold cross-validation experiments.The results confirm that the model exhibits strong generalization capabilities and can be applied to various scenarios.Moreover,the integration of the attention mechanism further enhances the predictive performance of the DNN model by mitigating the influence of other objective factors.展开更多
基金supported by the Liaoning Province Natural Science Foundation Project,China(2021-MS-238)the Scientific Research Project of Liaoning Provincial Department of Education,China(LJGD2020002)the Liaoyang Science and Technology Plan Project,China([2021]No.24-9).
文摘This research presents a novel approach that utilizes Deep Neural Network(DNN)technology to estimate glucose levels in enzyme-catalyzed oxidation reactions.Typically,glucose content is measured using glucose oxidase in biosensors.However,when this enzyme is used as a catalyst during the reaction,it may interfere with accuracy of the detection.To address this challenge,we propose a DNN-based method that incorporates an attention mechanism to improve glucose prediction accuracy.The study investigates the synthesis of zinc gluconate through enzyme-catalyzed processes,employing numerical simulations to analyze heat release and transfer in the reactor.By integrating the attention mechanism,the research enhances the predictive performance for glucose levels in these reactions.DNN models are known for their ability to learn complex patterns in data,and their capacity to handle large datasets makes them particularly suitable for predicting glucose levels in biochemical processes.This capability enables the model to effectively capture subtle relationships between variables.Experimental results demonstrate a significant improvement in predictive accuracy,with a mean absolute error(MAE)as low as 0.024,which is considerably better than conventional methods.The study also uses numerical simulations to analyze heat release and transfer during the reaction process and validates the model’s generalization ability through K-fold cross-validation experiments.The results confirm that the model exhibits strong generalization capabilities and can be applied to various scenarios.Moreover,the integration of the attention mechanism further enhances the predictive performance of the DNN model by mitigating the influence of other objective factors.
