Related Experiment Video
Updated: Aug 22, 2025
12:20
Use of the Operant Orofacial Pain Assessment Device OPAD to Measure Changes in Nociceptive Behavior
Published on: June 10, 2013
21.3K
Improving Ovine Behavioral Pain Diagnosis by Implementing Statistical Weightings Based on Logistic Regression and
Pedro Henrique Esteves Trindade1, João Fernando Serrajordia Rocha de Mello2, Nuno Emanuel Oliveira Figueiredo Silva1
1Department of Veterinary Surgery and Animal Reproduction, School of Veterinary Medicine and Animal Science, São Paulo State University, Botucatu 05508-270, SP, Brazil.
Animals : an Open Access Journal From MDPI
|November 11, 2022
Summary
Machine learning algorithms enhance the Unesp-Botucatu sheep acute pain scale (USAPS) by statistically weighting behavioral items. This improved pain diagnosis in sheep undergoing surgery.
Area of Science:
- Veterinary Medicine
- Animal Welfare
- Machine Learning in Animal Health
Background:
- The Unesp-Botucatu sheep acute pain scale (USAPS) is a validated tool for assessing pain in sheep.
- Statistical weighting of behavioral items can improve diagnostic instrument accuracy, but this has not been explored in animal pain scales.
- Accurate pain assessment is crucial for timely and effective analgesia in animals.
Purpose of the Study:
- To determine if implementing statistical weightings using machine learning algorithms enhances the discriminatory capacity of the USAPS.
- To compare the diagnostic accuracy of the original USAPS with machine learning-weighted versions.
Main Methods:
- Utilized a pre-existing behavioral database of 48 sheep in the perioperative period of laparoscopy.
- Applied a multilevel binomial logistic regression algorithm and a random forest algorithm to determine statistical weights.
- Classified sheep based on their need for analgesia and compared classification quality using the area under the curve (AUC).
Main Results:
- Both machine learning algorithms significantly improved the USAPS's discriminatory ability.
- The weighted USAPS using multilevel binomial logistic regression achieved an AUC of 96.59 (CI: [95.02-98.15]).
- The weighted USAPS using random forest achieved an AUC of 96.28 (CI: [94.17-97.85]), both higher than the original USAPS AUC of 94.87 (CI: [92.94-96.80]).
Conclusions:
- Implementation of statistical weights via machine learning algorithms demonstrably improves the discriminatory ability of the USAPS.
- Machine learning offers a powerful approach to refine animal pain assessment tools.
- These findings support the use of advanced algorithms for more precise pain diagnosis in veterinary medicine.
