Monitoring of Respiratory Disease Patterns in a Multimicrobially Infected Pig Population Using Artificial Intelligence and Aggregate Samples

Abstract

A 24/7 AI sound-based coughing monitoring system was applied in combination with oral fluids (OFs) and bioaerosol (AS)-based screening for respiratory pathogens in a conventional pig nursery. The objective was to assess the additional value of the AI to identify disease patterns in association with molecular diagnostics to gain information on the etiology of respiratory distress in a multimicrobially infected pig population. Respiratory distress was measured 24/7 by the AI and compared to human observations. Screening for swine influenza A virus (swIAV), porcine reproductive and respiratory disease virus (PRRSV), Mycoplasma (M.) hyopneumoniae, Actinobacillus (A.) pleuropneumoniae, and porcine circovirus 2 (PCV2) was conducted using qPCR. Except for M. hyopneumoniae, all of the investigated pathogens were detected within the study period. High swIAV-RNA loads in OFs and AS were significantly associated with a decrease in respiratory health, expressed by a respiratory health score calculated by the AI The odds of detecting PRRSV or A. pleuropneumoniae were significantly higher for OFs compared to AS. qPCR examinations of OFs revealed significantly lower Ct-values for swIAV and A. pleuropneumoniae compared to AS. In addition to acting as an early warning system, AI gained respiratory health data combined with laboratory diagnostics, can indicate the etiology of respiratory distress.

Keywords

• surveillance
• sample types
• novel
• diagnostics
• respiratory disease
• influenza