Koç Erdinç, Türkoğlu Muammer
Department of Business Administration, Faculty of Administrative Sciences and Economics, Bingol University, Bingol, Turkey.
Department of Computer Engineering, Faculty of Engineering and Architecture, Bingol University, Bingol, Turkey.
Signal Image Video Process. 2022;16(3):613-621. doi: 10.1007/s11760-020-01847-5. Epub 2021 Jan 25.
The need for healthcare equipment has increased due to the COVID-19 outbreak. Forecasting of these demands allows states to use their resources effectively. Artificial intelligence-based forecasting models play an important role in the forecasting of medical equipment demand during infectious disease periods. In this study, a deep model approach is presented, which is based on a multilayer long short-term memory network for forecasting of medical equipment demand and outbreak spreading, during the coronavirus outbreak (COVID-19). The proposed model consists of stages: normalization, deep LSTM networks and dropout-dense-regression layers, in order of process. Firstly, the daily input data were subjected to a normalization process. Afterward, the multilayer LSTM network model, which was a deep learning approach, was created and then fed into a dropout layer and a fully connected layer. Finally, the weights of the trained model were used to predict medical equipment demand and outbreak spreading in the following days. In experimental studies, 77-day COVID-19 data collected from the statistics data put together in Turkey were used. In order to test the proposed system, the data belonging to last 9 days of this data set were used and the performance of the proposed system was calculated using statistical algorithms, MAPE and . As a result of the experiments carried out, it was observed that the proposed model could be used to estimate the number of cases and medical equipment demand in the future in relation to COVID-19 disease.
由于新冠疫情爆发，对医疗设备的需求增加。对这些需求进行预测能使各国有效利用其资源。基于人工智能的预测模型在传染病期间医疗设备需求预测中发挥着重要作用。在本研究中，提出了一种深度模型方法，该方法基于多层长短期记忆网络，用于在新冠疫情（COVID-19）期间预测医疗设备需求和疫情传播。所提出的模型按处理顺序由以下阶段组成：归一化、深度长短期记忆网络以及随机失活-全连接-回归层。首先，对每日输入数据进行归一化处理。之后，创建作为深度学习方法的多层长短期记忆网络模型，然后将其输入到随机失活层和全连接层。最后，使用训练模型的权重来预测未来几天的医疗设备需求和疫情传播。在实验研究中，使用了从土耳其汇总的统计数据中收集的77天新冠疫情数据。为了测试所提出的系统，使用了该数据集最后9天的数据，并使用统计算法平均绝对百分比误差（MAPE）等来计算所提出系统的性能。实验结果表明，所提出的模型可用于估计未来与新冠疾病相关的病例数和医疗设备需求。
