Seasonal Inactivation of Cryptosporidium parvum Oocysts in Soil and Manure Microenvironments Using the LSTM-based...

Cryptosporidium parvum is a resilient zoonotic parasite with a high potential for environmental persistence, particularly in agricultural soils and manures. This study introduces a novel application of Long Short-Term Memory (LSTM) deep learning, an artificial intelligence model, to simulate realistic seasonal diurnal temperature and relative humidity cycles based on historical climate data from Baton Rouge, Louisiana, USA. The predicted conditions for summer (21-42 °C; 34-96% RH) and winter (1-18 °C; 50-90% RH) were applied in a controlled growth chamber to study oocyst inactivation over 30 days in soil and manure microenvironments. Inactivation followed first-order kinetics, with significantly higher decay rates observed in manure under summer conditions (k = -0.01379 day^-1) compared to winter (k = -0.00405 day^-1). Soil showed consistently slower inactivation rates than manure across both seasons. ANOVA and posthoc analyses confirmed the significance of temperature, substrate type, and their interaction on oocyst decay (p < 0.05). These findings highlight the critical influence of temperature and substrate properties on oocyst persistence and underscore the potential of LSTM-based climate modeling to improve environmental pathogen risk assessments under dynamic seasonal conditions.