融合定性定量分析和神经网络的失眠辨证辅助决策模型构建与验证OA

Objective To provide reference for clinical auxiliary diagnosis and application,medical case records of famous old traditional Chinese medicine(TCM)practitioners treating insomnia was explored based on TCM theory and neural network.Methods Based on 1205 cases of insomnia treated by famous old TCM practitioners,data standardization and coding were carried out.Extreme gradient boosting(XGBoost)algorithm quantitatively extracted the core symptoms and qualitatively analyzed their accuracy based on TCM theories.Insomnia syndrome differentiation assisted decision-making model was constructed based on the Backpropagation(BP)neural network architecture and integrated with the Particle swarm optimization(PSO).The model performance was evaluated by accuracy,mean square error(MSE)and F1-score,and the reliability of the model was verified by simulation data.Results 73 core symptoms were screened out by comprehensive XGBoost feature extraction and TCM theory analysis.The accuracy of the optimized model was 88.80%,the MSE was 0.041,and the F1-scores of syndrome types 1-5 were 89.86%,87.76%,93.22%,87.03%,84.51%,respectively.And the accuracy of simulation data validation was 89.17%.Conclusion The fusion of qualitative theory and quantitative data enhances the interpretability of the feature extraction process,and the assisted decision-making model for insomnia syndrome differentiation constructed by fusing qualitative and quantitative analysis has good performance,providing a new method for intelligent diagnosis and treatment of TCM.