Author : Emil A. Walleser
Publisher :
ISBN 13 :
Total Pages : 0 pages
Book Rating : 4.:/5 (139 download)
Book Synopsis Applied Technology for Disease Prediction and Health Management in Dairy Cattle by : Emil A. Walleser
Download or read book Applied Technology for Disease Prediction and Health Management in Dairy Cattle written by Emil A. Walleser and published by . This book was released on 2023 with total page 0 pages. Available in PDF, EPUB and Kindle. Book excerpt: Dairy cows undergo enormous metabolic changes following parturition and with the onset of lactation. Failure to execute these adaptations results in disease, reduced production, and loss of animal welfare. This period has been a major focus of research for the past twenty years. At the same time, the dairy industry has undergone a data revolution. Advancements in computational tools, predictive modeling, and data analytics tools offer new opportunities for early disease diagnosis and monitoring. This dissertation explores applications of data tools for monitoring and predicting disease with a primary focus on the transition period. Specific technologies investigated are milk analysis for prediction of transition period health disorders, herd data analysis platforms, and nanopore technology for infectious disease management. Milk analysis is common on dairies for monitoring production and milk composition. Infrared spectroscopy is applied to a milk sample and the resulting data can be used to determine milk components such as fat and protein, blood biomarkers, and other expressed phenotypes. The first objective of chapter 2 was to develop predictive models for blood beta-hydroxybutyrate (BHB) and blood nonesterified fatty acids (NEFA) using milk Fourier-transform infrared spectroscopy (FTIR) data and known blood levels. The second objective was to categorize cows as either metabolically disordered or healthy using a combination of blood BHB, blood NEFA, and milk fat protein quotient. We demonstrated screening level diagnostic performance for milk FTIR prediction of BHB and NEFA and reasonable performance for classification of metabolic disturbance. In chapter 3 we developed novel prediction models for hyperketonemia prediction using milk FTIR data. We collected a large and diverse group of milk FTIR samples from multiple locations, breeds, and lactations. We hypothesized that applying deep learning and ensembling or averaging model results together would improve performance compared to conventional predictive algorithms. We then implemented and evaluated multiple algorithms for their performance on hyperketonemia classification. Predictive performance was similar across modeling options and performed at screening test level. We demonstrated prediction models can be developed using more diverse groups of cattle than previously evaluated. While BHB and hyperketonemia have been a key research area in the transition period for many years, the effect of inflammation on transition cow health is becoming a key topic. However, the lack of cowside testing tools makes routine monitoring of inflammation levels extremely challenging. Milk FTIR-based predictions are a possible tool for monitoring inflammation on a cow and herd level. Multiple biomarkers of inflammation exist, with some of the most commonly measured being the acute phase proteins. Acute phase proteins are either elevated or reduced in response to inflammatory stimuli, haptoglobin is one of the most commonly measured due to its high peak levels and long period of elevation. In chapter 4 we developed regression and classification models for blood and milk haptoglobin models. Predictive performance was limited for both blood and milk haptoglobin suggesting inflammatory monitoring using milk FTIR may be better utilized on a herd-level. Applications of data analysis on the dairy extend beyond sample analysis into farm records. Large scale data analysis and predictive modeling are the future of cow management, but the tools of today limit the usability of on-farm data. Dairy record tools limit the ability to perform herd health analysis and improve cow health. In chapter 5 we implemented and provided a case study for Dairy-Copilot, reducing time to inference from hours to minutes. We believe providing user-accessible tools will drive improved cow health and producer profitability simultaneously. In chapter 6, we look to the future, nanopore sequencing allows real-time sequencing for samples using only a handheld sequencer and laptop computer. We conducted a proof-of-concept project to investigate predicting gene sequences using convolutional neural networks. In the second part we generated a synthetic metagenomic dataset to predict salmonella genome sequences vs other bacterial species. Our work demonstrates that convolutional neural networks are capable of predicting short conserved sequences, but prediction of salmonella sequences in metagenomic samples requires more investigation. The final chapter contains thoughts on the future extending beyond literature. We explore the numerous opportunities data brings to the dairy but caution that without proper oversight many negatives exist. Finally we remark on the dairy cow, the surrogate mother of humanity, calling for her continued care and respect.
