Related Experiment Video
Updated: Jun 5, 2025
Machine Learning Algorithms for Early Detection of Bone Metastases in an Experimental Rat Model
Published on: August 16, 2020
Data-driven insights into pre-slaughter mortality: Machine learning for predicting high dead on arrival in meat-type
Chalita Jainonthee1, Phutsadee Sanwisate2, Panneepa Sivapirunthep3
1PhD Program in Veterinary Science (International Program), Faculty of Veterinary Medicine, Chiang Mai University, under the CMU Presidential Scholarship; Veterinary Public Health and Food Safety Centre for Asia Pacific (VPHCAP), Faculty of Veterinary Medicine, Chiang Mai University, Chiang Mai 50100, Thailand; Research Center for Veterinary Biosciences and Veterinary Public Health, Faculty of Veterinary Medicine, Chiang Mai University, Chiang Mai 50100, Thailand.
High dead on arrival (DOA) rates in meat-type ducks are influenced by factors like truckload size and lairage temperature. Machine learning models identified key predictors to minimize pre-slaughter losses and improve animal welfare.
Area of Science:
- Animal Science
- Agricultural Economics
- Veterinary Public Health
Background:
- Dead on arrival (DOA) in poultry signifies welfare issues and economic losses.
- High DOA rates can result from various pre-slaughter conditions.
- Understanding contributing factors is crucial for mitigation strategies.
Purpose of the Study:
- To classify high DOA rates (≥ 0.15%) in meat-type ducks using machine learning.
- To identify key predictors influencing DOA rates during pre-slaughter transport.
- To analyze variable importance for predicting and minimizing DOA.
Main Methods:
- Utilized machine learning algorithms (LASSO, SVM, DT, RF, XGBoost) on 18,643 truckload entries.
- Employed data-sampling techniques (oversampling, undersampling, ROSE, SMOTE) for imbalanced data.
- Analyzed predictors: season, time, load size, distance, duration, age, weight, lairage time, and temperature.
Main Results:
- XGBoost-Up, XGBoost-Down, and RF-Down models demonstrated superior performance in classifying high DOA.
- Key predictors for high DOA were number of ducks per truckload, lairage temperature, and average body weight.
- Secondary factors included transportation duration, distance, and period.
Conclusions:
- Machine learning effectively classifies high DOA rates in meat-type ducks.
- Managing truckload size, lairage temperature, and body weight can reduce pre-slaughter mortality.
- Further investigation and management of identified factors are recommended to minimize economic losses and enhance animal welfare.
