Machine Learning-Based Characterization of Acute Leptospira spp. Infections in Dogs: Insights from a Biological Isolation Unit Study

Abstract

Leptospirosis is a zoonotic disease caused by pathogenic Leptospira species, posing a significant threat to both animal and human health worldwide. While traditional diagnostic methods rely on serological tests and polymerase chain reaction (PCR), there's a growing interest in leveraging machine learning (ML) algorithms for disease characterization. In this study, we present insights gained from employing ML techniques to characterize acute Leptospira spp. infections in dogs, utilizing data from a Biological Isolation Unit.Data collection involved clinical and laboratory parameters from 150 dogs diagnosed with acute leptospirosis. Various ML algorithms, including Random Forest, Support Vector Machines, and Neural Networks, were trained and validated on the dataset to discern patterns indicative of acute infection. Feature selection techniques were employed to identify the most influential parameters contributing to diagnostic accuracy.Results demonstrated the efficacy of ML models in accurately distinguishing acute Leptospira spp. infections in dogs. Random Forest exhibited the highest performance metrics, achieving an accuracy of 92% and an area under the receiver operating characteristic curve (AUC-ROC) of 0.95. Key features contributing to classification included clinical signs such as fever, vomiting, and icterus, as well as laboratory parameters like elevated liver enzymes and thrombocytopenia.Furthermore, the ML approach provided insights into complex interactions among variables, shedding light on potential biomarkers and underlying mechanisms of leptospiral pathogenesis. These findings underscore the potential of ML-based approaches in enhancing diagnostic accuracy and understanding the pathophysiology of infectious diseases, facilitating timely interventions and improved outcomes for both veterinary and public health.