A breathprint of influenza a virus infection in the ferret model

Influenza viruses (IV) pose a major public health concern, since they are highly contagious and still have a high global annual mortality. Current diagnostic methods are slow or have high error rates; therefore, a rapid detection tool is highly desirable to inform clinical management. Exhaled breath is a useful diagnostic in a number of diseases. In this work, we examined the exhaled breath of six ferrets collected pre- and post-IV infection. Exhaled breath analysis was carried out using a comprehensive two-dimensional gas chromatography (GC×GC) hyphenated with a time-of-flight mass spectrometer (ToF MS). Random Forest, a non-parametric machine learning algorithm, was used to select the most relevant diagnostic volatile organic compound (VOC) features and build the model using a cross-validation approach, obtaining an accuracy of 0.973 (out of 1.000). Using this approach, we defined 31 VOC features, which together produce a profile capable of discriminating between uninfected and IV-infected ferrets. Further characterization revealed an abundance of hydrocarbons, which is consistent with increased oxidative stress known to occur during viral infection. Our study is the first of its kind to define a unique exhaled breath signature for influenza infection in ferrets and will be useful for development of a rapid precision diagnostic technique