Use of machine-learning algorithms to aid in the early detection of leptospirosis in dogs

Reagan KL, Deng S, Sykes JE, et al. Use of machine-learning algorithms to aid in the early detection of leptospirosis in dogs. J Vet Diagn Invest. 2022;34(4). doi:10.1177/10406387221096781

Leptospirosis is a zoonotic disease with serious clinical effects, including renal, hepatic, and pulmonary complications in dogs. Traditional diagnostic methods, like the microscopic agglutination test (MAT), are limited by low sensitivity early in the disease process. This study applied machine-learning models (MLMs) using clinical data from 91 leptospirosis-positive and 322 negative dogs to predict infection with or without MAT results. The models demonstrated 100% sensitivity and up to 93.2% specificity, significantly improving early detection compared to serologic screening methods.