基于改进LSTM的多源绩效数据综合评估与预测模型设计

Design of multi-source performance data comprehensive evaluation and forecast model based on improved LSTM

传统绩效的评估与预测方法采取指标单一且覆盖面较窄,无法对医院绩效情况进行准确估计。针对上述不足,文中依托深度学习技术构建了多源绩效数据评估预测模型。根据DRG模型选取各项指标对传统LSTM网络进行改进,并采用双向网络提高了参数迭代的准确度。同时通过引入注意力机制有效提升了权重因子的计算精确度,再建立三重残差网络实现了对数据的训练与预测。数据实验结果表明,所提算法在预测拟合度方面与实际结果最为接近,且其误差指标在对比算法中均为最优,从而证明了该算法具有较好的综合性能。

The traditional performance evaluation and prediction methods adopt single index and narrow coverage,which can not accurately estimate the hospital performance.In view of the above shortcomings,a multi-source performance data evaluation and prediction model is constructed based on deep learning technology,selects various indicators according to the DRG model,improves the traditional LSTM network,and uses a two-way network to improve the accuracy of parameter iteration.By introducing the attention mechanism,the calculation accuracy of weight factors is effectively improved,and finally a triple residual network is established to train and predict the data.The experimental results show that the proposed algorithm is the closest to the actual results in terms of prediction fit,and its error index is the best in the comparison algorithm,which proves that the algorithm has good comprehensive performance.