论文部分内容阅读
重症监护病房中的病人身体状况通常很不稳定,常出现各种需要医护人员介入治疗的紧急状况。由于医疗资源有限,医护人员可能无法及时发现并处理这些紧急状况,给病人的存活率带来严重的负面影响。如果可以预测这些紧急状况的发生,并及时通知相关医护人员,将大大提高病人的存活率。常见重症监护病房紧急状况包括突然死亡、败血症、肺部感染、急性低血压、以及器官衰竭等。紧急状况预警建模主要采用病人的长时间生命体征监测数据,预测在一定时间之后发生某种紧急状况的可能性。预警模型所采用的监测数据分为静态数据、事件数据和时间序列数据等三类。静态数据具有容易采集、但预测准确性偏低的特点。事件数据或时间序列数据、以及多种类型数据的混合数据对于紧急状况预警模型的预测性能的提高有重要作用,将会获得更广泛的应用。
The patient’s condition in the intensive care unit is usually very unstable and there are often emergencies in which medical staff are required to intervene. Due to the limited medical resources, medical staff may not find and deal with these emergencies in a timely manner, which will have a serious negative impact on the patient’s survival rate. If you can predict the occurrence of these emergencies, and promptly notify the relevant medical staff, will greatly enhance the patient’s survival rate. Common ICU emergencies include sudden death, sepsis, lung infections, acute hypotension, and organ failure. Early warning modeling uses the patient’s long-term vital signs monitoring data to predict the likelihood of an emergency occurring after a certain period of time. The monitoring data used in the early warning model is divided into three categories: static data, event data and time series data. Static data has the characteristics of easy acquisition but low prediction accuracy. Incident data or time series data, as well as mixed data of many types of data, play an important role in predicting the performance of the emergency alert model and will be more widely used.