HospitalOPD Bottleneck Labelled dataset

The dataset contains 10,000 event records representing patient journeys through a hospital Outpatient Department (OPD). The dataset was developed to support research in healthcare process mining, workflow analytics, bottleneck detection, and predictive process monitoring. It captures the sequence of activities performed during patient visits, including consultations, laboratory investigations, medical imaging services, and pharmacy interactions. The dataset is structured as an event log where each patient visit is represented as a process instance and each healthcare activity is recorded as an event with associated timestamps. These timestamps enable the reconstruction of complete patient pathways and the calculation of workflow performance indicators such as waiting times, processing durations, throughput times, and delay propagation. In addition to traditional event log attributes, the dataset includes process mining features describing workflow behavior, resource utilization indicators reflecting workload conditions, and graph-based structural features derived from workflow graph analysis. Centrality measures such as degree, betweenness, closeness, eigenvector, and eccentricity were computed to quantify the structural importance of activities within the healthcare workflow network. A binary bottleneck label is also included, identifying events associated with workflow congestion and excessive delays. This allows the dataset to be used for supervised machine learning tasks related to predictive bottleneck identification. The dataset provides a comprehensive benchmark for studying patient flow dynamics, healthcare workflow optimization, process-aware machine learning, workflow graph analytics, and predictive process monitoring. By integrating temporal, behavioral, resource-related, and structural workflow characteristics, it enables researchers to investigate how workflow structure and operational performance jointly influence bottleneck formation in healthcare systems.本数据集包含10000条事件记录，刻画了患者在医院门诊科室（Outpatient Department, OPD）的就诊全流程。本数据集专为支持医疗流程挖掘、工作流分析、瓶颈检测以及预测性流程监控领域的研究而构建。 本数据集采用事件日志（event log）结构：每一次患者就诊对应一个流程实例，每一项医疗服务活动均作为带关联时间戳的事件进行记录。这些时间戳可用于完整复现患者就诊路径，并计算工作流性能指标，如等待时长、处理耗时、吞吐时长以及延迟传播情况。 除传统事件日志属性外，本数据集还包含刻画工作流行为的流程挖掘特征、反映工作负载状况的资源利用率指标，以及基于工作流图分析得到的图结构特征。研究人员计算了度（degree）、介数（betweenness）、接近度（closeness）、特征向量（eigenvector）与离心率（eccentricity）等中心性度量指标，以量化医疗工作流网络中各活动的结构重要性。 数据集还附带二元瓶颈标签，用于标记与工作流拥塞及过度延迟相关的事件，这使得本数据集可用于与预测性瓶颈识别相关的监督机器学习任务。 本数据集为研究患者流动动力学、医疗工作流优化、感知流程的机器学习、工作流图分析以及预测性流程监控提供了全面的基准支持。通过整合时间、行为、资源相关及结构层面的工作流特征，该数据集可帮助研究人员探究工作流结构与运营性能如何共同影响医疗系统中的瓶颈形成。