Europe Union

How our AI-driven algorithms enabled the detection of early signs of cows’ diseases

How our AI-driven algorithms enabled the detection of early signs of cows’ diseases
Is sustainable agriculture only a trend? For modern farmers, it’s necessary to maintain an advantage in the market and provide conscious consumers with the expected quality while maintaining competitive prices.

How can farmers decrease excessive antibiotic use while avoiding disease transmission?

Predictive AI algorithms empower farmers with timely insights into their herd’s health, enabling proactive intervention and minimizing economic losses. By detecting diseases early, farmers can implement targeted treatments and prevent the spread of illness within the herd. This proactive approach enhances farm effectiveness and reduces veterinary costs.

Early disease detection contributes to environmental sustainability by minimizing the need for excessive antibiotic use and reducing the environmental impact of disease outbreaks. By preventing the spread of illness, farmers can maintain a healthier herd, reducing the need for antimicrobial interventions and mitigating the risk of antibiotic resistance. 

Proactive disease management enhances food safety and traceability, fostering consumer confidence in the agricultural supply chain. Ultimately, prioritizing animal welfare translates into better products for customers, enhancing brand reputation and market competitiveness.

We took a challenge to use AI to improve animal welfare

In the case of cow breeding, modern farms utilize data from two sources: milking robots or cow collars to manage their livestock. Thanks to a predictive AI-based approach, we decided to use that information to help farmers maintain animal welfare.

Our experts collaborated with research institutions and farms in Latvia and the Czech Republic to address the complex challenge of collecting and analyzing diverse data sets related to cow health.  We designed a solution based on collected milk parameters from robotic milking robots in Latvia (mostly the information about fat/protein ratio) and biometric data from cow collars in the Czech Republic, including pH and temperature measured through fistula probes to detect early signs of diseases.

AI Animal Welfare Disease Detection Solution

How to deal with a large amount of information from various sensors and use it for animal health prediction?

Our experts faced several challenges, including integrating heterogeneous data sources, the need for a scalable and farm-agnostic data aggregation system, and the development of accurate and efficient AI models for health prediction. 

Each farm uses different devices, which can make analyzing inconsistent data challenging. Our experts’ architecture enables the integration of any sensors. As experts in Big Data, we have designed a scalable solution that can handle any number of data sources, making it suitable for small and large dairy farms.’s solution: AI-driven Livestock Health Monitoring’s AI expert, Marek Tatara, PhD, developed two AI algorithms for cows’ health monitoring:

  • An algorithm that utilized data on pH and temperature from cow collars to predict potential malnutrition issues (and therefore potential disease or poor feeding) with almost 100% accuracy,
  • An algorithm that analyzed milk composition, specifically the fat/protein ratio, to detect early signs of acidosis and ketosis.

Until now, information about the possibility of developing the disease was determined based on persistent alarming symptoms obtained by 12 readings taken every 15 minutes. Created by Marek Tatara, PhD, an artificial intelligence algorithm based on the data from cows’ collars, a precisely recurrent neural network, could predict whether a cow was sick based on temperature and pH. Thanks to our predictive algorithm, we have shortened detection time by three times. 

Our AI expert investigated the possibility of detecting early symptoms of acidosis and ketosis by analyzing milk composition and, more specifically, by testing the fat/protein ratio thanks to data obtained from milking robots. His algorithm enabled the disease to be forecasted at a very early stage.

AI-driven Livestock Health Monitoring Solution

We leveraged our IoT expertise to enable the AI-based algorithm to operate in real-time

Our Solutions Architect developed Steam Processing Engine, a robust infrastructure for aggregating sensor data from multiple farms. This system was designed to be agnostic to farm specifics, data types, and communication protocols, facilitating seamless data collection.

We have combined our experience in Big Data and AI to develop a comprehensive solution that detects anomalies at a very early stage and processes huge amounts of data collected from many sensors into useful information in real-time. Our system enables farmers to take actions that will prevent the development of diseases.

AI Early Detection System for Livestock Health

Our predictive algorithms enabled the detection of malnutrition in cows 3 times faster

The project achieved significant outcomes:

An algorithm for detecting health issues in cows with almost 100% accuracy significantly reduces the time needed to identify sick animals.

An algorithm for cows’ disease forecasting based on milk quality, enabling predictive maintenance of livestock.

The development of a scalable solution that can be adapted to various farm sizes and types, potentially transforming livestock management practices.

An improvement in animal welfare through early disease detection and intervention, reducing the overall impact of diseases on livestock productivity.

Technology stack

  • Kafka for efficient data streaming.
  • MQTT protocol for lightweight messaging across networked devices.
  • Python and TensorFlow for the development and training of AI models.
  • Java/Spring boot for microservices development.
  • Kubernetes and Helm for infrastructure management and monitoring.
kafka icon
TensorFlow icon
Python icon
spring icon
kubernetes icon

HELM icon

Java icon
AI Temperature Monitoring in Cattle

Read More of Our Case Studies.

Let’s connect!

Send us an e-mail: [email protected]