NODES - SPOKE 6 Project FORMIDABILÆ - D4.2 Simulation model for heat stress scenarios

This document focused on predicting milk yield and monitoring heat stress in dairy cows using data-driven and machine-learning approaches. Data from commercial farms in Italy were collected during summer 2022, including daily milk yield, parity, lactation stage, milking information, feed intake, and environmental data (temperature, humidity, THI). Several prediction models were tested: a General Linear Model (GLM) and machine learning algorithms (Random Forest, Gradient Boosting Machine, and Extreme Gradient Boosting). In internal validation, GBM achieved the highest accuracy, but in external validation on an independent farm, GLM performed best, showing the highest generalizability and the lowest prediction error. A deep learning forecasting system was developed using TimeSeries K-Means clustering and LSTM neural networks to predict next-day milk yield from the previous 7 days of data. Grouping cows by production patterns improved prediction accuracy and adaptability, especially during heat waves. Based on these results, Heat Guard, a decision-support application, was developed in both web (Shiny) and Android formats. The app automatically calculates heat-stress indices (THI, Adjusted THI, ETIC), classifies risk levels, and provides 5-day forecasts using real-time weather data. In addition, an interactive Heat Guard chatbot was developed using Zephyr-7B and semantic search to answer questions related to heat stress management. Overall, the outcomes demonstrate that combining predictive modeling, real-time monitoring tools, and AI-based decision support can improve management responses to heat stress and support farmers in maintaining milk production efficiency under future climate challenges. This document is a deliverable of the project FORMIDABILÆ ("Forage system to make resilient Maize, Dairy and Biogas supply chains for a Lasting Agricultural Ecosystem"), which promotes approaches aimed at improving the smart, resilient, circular and diversified farm to ensure food security and social economic sustainability of dairy food chains with the aim of reducing greenhouse gas emissions in this sector. This document is part of the project NODES which has received funding from the MUR – Missione 4, Componente 2, Investimento 1.5 – Creazione e rafforzamento di “Ecosistemi dell’innovazione”, costruzione di “leader territoriali di R&S'' – del PNRR funded by the European Union - NextGenerationEU with grant agreement no. ECS00000036