Native Microbials and Gutz Analytics have partnered to predict methane emissions in dairy cows using AI models based on next-generation sequencing of cow saliva. The partnership aims to fill a gap in greenhouse gas-related sustainability efforts, which is the development of high-throughput, cost-effective, and non-invasive tools for measuring and predicting enteric methane emissions in cattle. Using data from approximately 2,500 rumen and saliva samples collected from 142 cows, AI models were trained to focus on microbes directly or indirectly involved in methane production.
The training process leveraged recurrent neural networks to track microbiome changes over time and incorporated covariates like feed intake and milk production efficiency, which are linked to methane emissions. Despite challenges from data variability and complex sequencing data, the developed model has shown a 5-fold reduction in prediction error compared to traditional big data models. Overall, this saliva-based method of estimating methane emissions, which is implemented as a simple ordinary least-squares regression model, is comparable to current methane measurement approaches (R2= 0.90) and is expected to improve as additional validation is performed.
Read more here: https://gutzanalytics.com/native-case-study
Comments