Data for: Combining Time Series and Recurrent Network Approaches for Long-Term Prediction of Emergency Department Admissions

Admissions to emergency departments exhibit high variability. This complicates the adjustment of resource allocation, resulting in rigid policies that lead to inefficient service. A number of studies have approached this issue by employing statistical analysis, machine learning regressors, or time series data models. We propose the use of improved recurrent neural networks that take into account the dynamic nature of the data and experience, introducing contextual variables that lead to better predictability. This allows the novelty of formulating a solution to a requirement that hospital managers have expressed in regard to the planning of available resources, a need that has thus far gone unmet: the availability of long prediction horizons for the purpose of long-term resource allocation. The present dataset contains data obtained from a medical facility. The formulation of the problem and detailed description of the dataset can be found in the article "Combining Time Series and Recurrent Network Approaches for Long-Term Prediction of Emergency Department Admissions".

急诊科室的入院量呈现高度变异性。这一特性使得资源调配的调整工作复杂化，进而催生了僵化的运营策略，最终导致服务效率低下。已有多项研究通过统计分析、机器学习回归模型及时序数据模型等手段对该问题进行了探索。本文提出采用改进型循环神经网络（Recurrent Neural Network），该模型能够考量数据的动态特性与实际场景经验，并引入上下文变量以提升预测性能。该方案创新性地构建了解决方案，以回应医院管理者在规划可用医疗资源时提出的一项迄今尚未得到满足的核心需求：即获取长预测时域，以支撑长期资源调配工作。本数据集包含从某医疗机构采集的相关数据。该问题的建模思路与数据集的详细说明均可在题为"Combining Time Series and Recurrent Network Approaches for Long-Term Prediction of Emergency Department Admissions"的学术论文中查阅。