老年人多病共存非常常见。老年共病特点及衍变的研究对患者自身管理、诊疗质量改善和医疗资源配置有着重大意义。既往临床研究和指南多关注单个疾病或少量伴随疾病。本项目拟通过对国家医疗数据中心的四百多家三甲医院上千万病人诊疗信息及北京大学附属综合医院电子病历信息的有机整合与分析挖掘，结合医学信息学信息采集利用与流行病学建模方法，挖掘中国老年人共病模式，分析共病发展规律及交互作用对临床结局的影响，构建基于疾病交互作用的临床转归预测模型，有重要的理论及应用价值。研究方法包括：判别分析构建疾病分类模型；关联分析等无监督模型探索共存疾病间累加效应；压缩工具模型筛选诊疗模式及广义线性模型确立权重；建立随访队列，利用竞争风险模型和随机效应模型预测重返和急诊入院事件。研究探讨老年人多病共存诊疗模式及临床转归预测模型可辅助医生以病人为中心制定整体性和个性化诊疗方案，为老年病人分级诊疗提供依据。

Multimorbidity among older Chinese has been a common problem. Study on the characteristics and evolution of the multimorbidity among elderly is of great significance on the patient's self-management, improvement of treatment, and medical resource allocation. However, clinical studies have largely focused on individual diseases or a few concomitant diseases. Using clinical core information of over tens of millions of inpatients from hundreds of hospitals throughout China and Electronic Medical Records from Peking University affiliated hospitals, we propose to utilize the methodology of medical informatics and epidemiology to study multimorbidity patterns、disease trajectories and clinical outcomes, and finally to develop the risk prediction models by data integration and mining techniques. Our research methods will include discriminant analysis to construct disease classification models, association analysis to explore addictive effects between multiple conditions, shrinkage methods to select multimorbidity patterns, generalized linear models to determine the weights, as well as competing risk models and random effect models to predict the risk of hospital readmission and emergency admission. Studying multimorbidity pattern and optimal risk prediction models of clinical outcomes will be helpful to assist physicians to develop the holistic and individualized treatment plan, and provide evidence for the classification of diagnosis and treatment among older patients.