Using Artificial Intelligence to Predict Survivability Likelihood and Need for Surgery in Horses Presented With Acute...

Area of Science:

• Veterinary Medicine
• Artificial Intelligence
• Machine Learning

Background:

• AI and machine learning applications are advancing rapidly in human medicine but are underutilized in veterinary fields.
• The veterinary field presents numerous opportunities for AI and machine learning to improve diagnostic and prognostic capabilities.

Purpose of the Study:

• To investigate the efficacy of AI and machine learning algorithms in predicting surgical intervention and survival likelihood in horses experiencing acute abdominal pain (colic).
• To evaluate the performance of Decision Trees, Multilayer Perceptron, Bayes Network, and Naïve Bayes algorithms in this veterinary context.

Main Methods:

• Utilized clinical data, including patient history, physical examination findings, and diagnostic procedures, from horses presenting with colic.
• Applied four distinct machine learning algorithms: Decision Trees, Multilayer Perceptron, Bayes Network, and Naïve Bayes.

Main Results:

• Machine learning algorithms achieved 76% accuracy in predicting the necessity for surgical intervention in horses with colic.
• The algorithms demonstrated 85% accuracy in predicting the likelihood of survival for affected horses.

Conclusions:

• AI and machine learning models can effectively predict critical outcomes in equine colic cases.
• The findings support the value and potential of applying AI and machine learning technologies across various veterinary clinical applications, warranting further research.