An Observational Clinical Study on the Construction of an Artificial Neural Network Model for ICU Pneumonia

An Observational Clinical Study on the Construction of an Artificial Neural Network Model for Rapid Intelligent Diagnosis of Microorganisms in ICU Pneumonia Based on Species-specific Rapid Detection of Pathogenic Bacteria

To achieve rapid, intelligent and accurate microbiological diagnosis and treatment for ICU pneumonia, an artificial neural network model for microbiological diagnosis is established, which depends on many clinical cases and machine deep learning from clinical experts' judgements according to species-specific rapid detection of pathogenic bacteria and other clinical parameter variables of patients.

This study is a prospective single-centre observational study, 600 ICU pneumonia patients are expected to be selected as the observation object, and the lower respiratory secretions of patients on d1, d3 and d7 after enrollment are collected for species-specific rapid detection and microbial culture, while the general information of the patients and the clinical information of the corresponding time points on d1, d3 and d7 are collected. Two experienced senior physicians were organized to determine whether the microbial results were colonized or infected, and an artificial neural network model for rapid and intelligent diagnosis of pathogenic microorganisms in ICU pneumonia will be established and validated through multi-dimensional machine learning.

Inclusion Criteria:

1. aged ≥18 years;
2. agreed to obtain lower respiratory specimens for rapid testing of pathogenic bacteria;
3. all were enrolled by an experienced physician who dynamically determined that the microorganisms were in a colonised or infected state;
4. signed an informed consent form.

Exclusion Criteria:

1. pregnant women;
2. lactating women;
3. patients who could not obtain lower respiratory specimens;