On-Site Diagnosis of Poultry Coccidiosis by a Miniature Mass Spectrometer and Machine Learning

Abstract

Coccidiosis is a prevalent parasitic disease in poultry that costs the modern poultry industry more than US$3 billion worldwide every year for the resulting prevention, treatment, and related production loss. Currently, the diagnosis of coccidiosis relies on detecting the oocysts of Eimeria, a genus that causes coccidiosis, in feces by a veterinarian or polymerase chain reaction. However, both methods must be performed in a laboratory and could observe the oocysts of Eimeria only 4 days after infection. Herein, we present an on-site, early coccidiosis diagnosis platform by differentiating the fecal metabolic profile acquired using a miniature mass spectrometer between healthy and Eimeria tenella-infected chickens using a machine learning model. Our model achieved an accuracy of 94%, a sensitivity of 88%, and a specificity of 100%. It is noteworthy that our model can diagnose E. tenella infection 2 days earlier than current diagnostic methods. To make the mass spectrometry-based diagnosis platform usable in the field and feasible, a cart carrying all of the equipment and a user-friendly data processing graphical user interface was designed. On average, our diagnositc approach takes ∼9 min to obtain the results for each sample. This platform, miniature mass spectrometry-based metabolomics (M3S), provides a solution for early detection of parasite infection, from which an accurate, easy-to-determine on-site coccidiosis diagnosis can be realized.