Pain recognition is fundamental for safeguarding animal welfare. Facial expressions have been investigated in several species and grimace scales have been developed as pain assessment tool in many species including horses (HGS) and mice (MGS). This study is intended to progress the validation of grimace scales, by proposing a statistical approach to identify a classifier that can estimate the pain status of the animal based on Facial Action Units (FAUs) included in HGS and MGS. To achieve this aim, through a validity study, the relation between FAUs included in HGS and MGS and the real pain condition was investigated. A specific statistical approach (Cumulative Link Mixed Model, Inter-rater reliability, Multiple Correspondence Analysis, Linear Discriminant Analysis and Support Vector Machines) was applied to two datasets. Our results confirm the reliability of both scales and show that individual FAU scores of HGS and MGS are related to the pain state of the animal. Finally, we identified the optimal weights of the FAU scores that can be used to best classify animals in pain with an accuracy greater than 70%. For the first time, this study describes a statistical approach to develop a classifier, based on HGS and MGS, for estimating the pain status of animals. The classifier proposed is the starting point to develop a computer-based image analysis for the automatic recognition of pain in horses and mice.

Can grimace scales estimate the pain status in horses and mice? A statistical approach to identify a classifier

PASCUZZO, RICCARDO;Vantini, Simone;
2018-01-01

Abstract

Pain recognition is fundamental for safeguarding animal welfare. Facial expressions have been investigated in several species and grimace scales have been developed as pain assessment tool in many species including horses (HGS) and mice (MGS). This study is intended to progress the validation of grimace scales, by proposing a statistical approach to identify a classifier that can estimate the pain status of the animal based on Facial Action Units (FAUs) included in HGS and MGS. To achieve this aim, through a validity study, the relation between FAUs included in HGS and MGS and the real pain condition was investigated. A specific statistical approach (Cumulative Link Mixed Model, Inter-rater reliability, Multiple Correspondence Analysis, Linear Discriminant Analysis and Support Vector Machines) was applied to two datasets. Our results confirm the reliability of both scales and show that individual FAU scores of HGS and MGS are related to the pain state of the animal. Finally, we identified the optimal weights of the FAU scores that can be used to best classify animals in pain with an accuracy greater than 70%. For the first time, this study describes a statistical approach to develop a classifier, based on HGS and MGS, for estimating the pain status of animals. The classifier proposed is the starting point to develop a computer-based image analysis for the automatic recognition of pain in horses and mice.
2018
File in questo prodotto:
File Dimensione Formato  
Dalla Costa et al 2018.pdf

accesso aperto

: Publisher’s version
Dimensione 2.99 MB
Formato Adobe PDF
2.99 MB Adobe PDF Visualizza/Apri

I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.

Utilizza questo identificativo per citare o creare un link a questo documento: https://hdl.handle.net/11311/1061417
Citazioni
  • ???jsp.display-item.citation.pmc??? 17
  • Scopus 36
  • ???jsp.display-item.citation.isi??? 32
social impact