Retrospective observational study to develop a Machine Learning Algorithm to evaluate parameters collected from routine data for the diagnosis of sepsis and septic shock and their influence on time to diagnosis and patient outcome.

Retrospective routine data from the medical records of the department of anesthesiology and operative intensive care from 01. 01. 2007 to 31. 12. 2021 are analyzed in digital form.

The first step is the development of a machine learning algorithm (MLA). This MLA will be validated and analyzed for his predictive value with regard to early diagnosis of sepsis/septic shock depending on the conceptual value of detection variables (Sepsis-3 vs. SIRS). Further analysis will focus on improvement of accuracy for the MLA and the effect of these detection variables on quality of treatment processes and also on economic consequences like cost and revenue.

Timeline:

1. Conception and development of the ML Algorithm (6 months)
2. Identification and diagnostic validation of sepsis patients (6 months)
3. Secondary analyses (36 months)

Inclusion Criteria:

• age >= 18 years
• ICU stay of > 24 hours

Exclusion Criteria:

• none

Intensive care unit patients (female, male, miscellaneous) from the age of 18 years with an intensive care stay of > 24 hours in an intensive care unit since 01. 01. 2007 from the Department of Anesthesiology and Operative Intensive Care Medicine (CCM/CVK)