In recent years, although the clinical treatment of sepsis has been greatly improved, it is still an important cause of death in ICU patients, and seriously threatens human health. Its predictive biomarkers have become one of the bottlenecks in the field of disease diagnosis, treatment and development of effective drugs to reduce incidence rate and mortality. This will eventually become the key point of treatment for patients with sepsis. In the early stage, the investigators have established a single center sepsis database and sepsis animal model, and made a preliminary exploration on the mechanism and treatment of sepsis. Based on the previous results, this study intends to create a national multi center sepsis apparent database and sample bank, collect the data of sepsis patients' injury characteristics, clinical characteristics, biochemical indicators, micro multidimensional and omics results, etiological characteristics, etc., and integrate them. Using big data combined with machine learning method, the early warning and real-time course monitoring model of traumatic sepsis is established. The completion of this project can achieve early warning of sepsis, real-time monitoring of the progress of the disease, early rational allocation of medical care, and reduce the mortality of sepsis patients.

Sepsis is one of the most fatal diseases worldwide, characterized by high incidence rate (18.6/1 000 hospitalization) and high mortality (50%). The patients often need to be treated in ICU, and the medical cost accounts for a large proportion. The ideal state, that is, accurate and early identification, must be the key point to influence the clinical decision-making of sepsis and guide more accurate treatment and intervention. With the development and improvement of pre hospital emergency technology, surgical technology and intensive care technology, the early mortality of patients with sepsis decreased significantly, but the mortality caused by multiple organ dysfunction (MODS) increased significantly. However, there are few reports on early sepsis warning and real-time monitoring of sepsis patients.

The existing research on early warning and course monitoring of sepsis can be roughly divided into demographic data, trauma severity score system, physiological and biochemical indicators, genetic background and so on. However, most studies only focus on the significance of a single index in the early warning and diagnosis of sepsis, which can only reflect one aspect of the body, and the diagnostic sensitivity is not high. Although there are a few multi marker related studies, such as the haplotype (- 1082-819-592ata) of three gene polymorphisms in IL-10 promoter region can affect the risk of sepsis in a small population (114 cases). The combination of plasma and cell biomarkers in critically ill patients suggests that the combination of plasma PCT, sTREM-1 and neutrophil CD64 index is better than single index in the early warning diagnosis of sepsis risk. However, this kind of research is still limited to a certain kind of indicators, and its clinical guidance value is limited. In addition, metabonomics and proteomics also have great potential to help identify specific sepsis phenotypes, and to find much-needed predictive and prognostic biomarkers, so as to guide more personalized management and treatment. Therefore, it is necessary to integrate the injury characteristics, clinical characteristics, biochemical indicators, micro multidimensional and omics results, etiological characteristics and other data to make accurate and efficient early warning and course monitoring of sepsis.

The project team has established a single center sepsis database in the early stage, and how to expand the scale of the database in the future, and use the samples in the sample library for multidimensional and omics methods to screen 100 biological molecular targets. Further research will integrate sepsis patients' injury characteristics, clinical characteristics, biochemical indicators, micro multidimensional and omics results, etiological characteristics and other relatively independent parts, and use big data combined with machine learning method to establish early warning and real-time course monitoring model of traumatic sepsis.

This study will be carried out from the following three levels: 1) to establish a multi center database of patients with sepsis; ② 100 biological molecular targets were screened by micro multidimensional and omics, and the data of injury characteristics, clinical characteristics, biochemical indexes, micro multidimensional and omics results, etiological characteristics and other aspects of sepsis patients were integrated to establish an early accurate early warning and real-time disease monitoring model of sepsis; ③ The application of the prediction model in sepsis patients was further verified by a cross regional multicenter prospective cohort study.

Inclusion Criteria:

• Patients diagnosed with an infectious disease or Sepsis patients meeting the diagnostic criteria for sepsis 3.0 (determined by two doctors with senior professional titles)
• Older than 16 years and younger than 60 years
• Agree to cooperate with the investigation and sign the informed consent

Exclusion Criteria:

• The infection was preexisting with clinically diagnosed organ insufficiency or failure
• Hematological disorders predate infection
• Sepsis is preceded by a serious infectious disease
• Long-term use of immunosuppressant or immunodeficiency patients.
• A pregnant woman

Patients with infectious inflammation or sepsis