Predicting Influenza A Virus Infection in the Lung from Hematological Data with Machine Learning.

The tracking of pathogen burden and host responses with minimally invasive methods during respiratory infections is central for monitoring disease development and guiding treatment decisions. Utilizing a standardized murine model of respiratory influenza A virus (IAV) infection, we developed and tested different supervised machine learning models to predict viral burden and immune response markers, i.e., cytokines and leukocytes in the lung, from hematological data. We performed independently infection experiments to acquire extensive data for training and testing of the models. We show here that lung viral load, neutrophil counts, cytokines (such as gamma interferon [IFN-γ] and interleukin 6 [IL-6]), and other lung infection markers can be predicted from hematological data. Furthermore, feature analysis of the models showed that blood granulocytes and platelets play a crucial role in prediction and are highly involved in the immune response against IAV. The proposed tools pave the path toward improved tracking and monitoring of influenza virus infections and possibly other respiratory infections based on minimally invasively obtained hematological parameters. During the course of respiratory infections such as influenza, we do have a very limited view of immunological indicators to objectively and quantitatively evaluate the outcome of a host. Methods for monitoring immunological markers in a host's lungs are invasive and expensive, and some of them are not feasible to perform. Using machine learning algorithms, we show for the first time that minimally invasively acquired hematological parameters can be used to infer lung viral burden, leukocytes, and cytokines following influenza virus infection in mice. The potential of the framework proposed here consists of a new qualitative vision of the disease processes in the lung compartment as a noninvasive tool.