| Popis: |
Too often ventilated patients at the intensive care unit (ICU) suffer from lung injury, lung infection, and complications which lead to a high mortality rate. Currently the diagnosis is based on chest imaging, clinical parameters, as well as microbiological indicators. It is often invasive and time-consuming, whereas timely treatment in the ICU is crucial. Breath analysis is a non-invasive method to help with clinical diagnosis and patient monitoring, especially in lung diseases. Many studies have shown the promising results in the respiratory field, e.g. lung cancer, asthma, chronic obstructive pulmonary disease (COPD). In this report, a methodology is described that enables the identification of biomarkers in Acute Respiratory Distress Syndrome (ARDS), based on the analysis of exhaled breath. The breath samples were collected from the ventilated patients at the ICU of the Academic Medical Center, in Amsterdam. The samples were analyzed off-line using a combination of Gas Chromatography and Mass Spectrometry (GC-MS), at the Philips Research laboratories. The features (ion-molecular fragments) of the GC-MS data were analyzed by sparse partial least squares (SPLS) and logistic regression, resulting in a classification model. The final model of 3 ion fragments all coming from a single molecule being octane, was tested successfully on an external validation set. The study has shown the feasibility of incorporating exhaled breath into ARDS diagnosis. The methodology we have developed is validated on a new set of samples, and can thus be extended to other diseases, e.g. Pneumonia. |
