PigAdvisor. Development and validation of a virtual advisor for decision-making in intensive pig farm management.
File: RTC-2017-5999-2 Modern animal production involves the generation of a massive amount of data or Big Data that needs more complex and complete management systems to optimize the use of it. This high volume of data requires new large-scale storage techniques and different approaches to retrieving information; the variety of data sources makes simple relational networks difficult to apply; and finally, the constant increase in the amount of data generated makes speed a key parameter in its handling. In a competitive environment such as that of livestock production, the primary objective must be to improve the efficiency of production systems, for which the correct management of data (collection, processing, analysis and distribution of information) generated every day on livestock farms is fundamental.
This project aims to optimise the productivity of intensive pig farms. To this end, a system for early detection of abnormal animal behaviour and sub-optimal environmental conditions will be generated based on the prediction algorithms developed and the specific production history of each farm. Likewise, an early diagnosis tool of the diseases will be offered through the online sending of the symptoms that have appeared in the farm and the telediagnosis. In other words, the scientific results obtained after the analysis of the databases available in this project will be applied in computer tools that will make it possible to apply the so-called Smart Farming. The main objective of the project is, therefore, the development of decision tools and automation technologies for the intelligent management of farms (Smart Farming), integrating different areas of knowledge to improve management, productivity and profits, making the most of the data obtained daily and which, to date, are not exploited. The system will generate suggestion algorithms based on the data obtained through image and sensor analysis that will help achieve behavioural and environmental alerts, as well as rapid syndromic telediagnosis and recommendations on the necessary analyses.
To obtain algorithms to predict reproductive results based on the feeding patterns of pregnant sows by correlating the data collected through feeding machines with the reproductive information contained in the farm management software.
To obtain algorithms to predict the incidence of disease as a function of the environmental conditions of the accommodation by correlating the data collected through sensors with the health information of the animals.
To generate a system of suggestion algorithms that allows a rapid syndromic telediagnosis based on the incorporation of the information contained in the databases of the clinical diagnostic laboratories participating in the project.
Development of a computer application capable of:
Collect the data collected through the feeding machines and environmental sensors, apply the algorithms developed in this project and generate health alerts for imminent risk of disease or low reproductive performance.
To send online images of syndromic manifestations and, based on the information contained in the software’s database, to offer a telediagnosis and the corresponding recommendations for laboratory analysis.
The project has been funded through the Challenge-Collaboration 2017 programme