Estimating military working dog core temperature change with machine learning: A simulation study.

Area of Science:

• Veterinary Medicine
• Machine Learning Applications
• Animal Physiology

Background:

• Military working dogs are vital for US military missions.
• Exposure to hot environments poses a significant hyperthermia risk, a leading cause of death in these dogs.
• Existing physics-based models estimate core temperature changes but may lack adaptability.

Purpose of the Study:

• To explore the feasibility of using machine learning (ML) to model military working dog core temperature changes.
• To develop an ML model that complements physics-based approaches by inferring real-world data relationships.
• To enhance model usability and applicability for atypical temperature patterns.

Main Methods:

• Approximated a physics-based model using three ML models.
• Developed and analyzed a random forest model, termed K9-TempML.
• Conducted feature analysis and augmentation studies on the K9-TempML model.

Main Results:

• Demonstrated the feasibility of applying ML to model canine core temperature changes.
• The K9-TempML model provided core temperature change estimates closest to the physics-based model.
• Feature analysis indicated that removing difficult-to-collect features could maintain accuracy while improving usability.

Conclusions:

• Machine learning offers a viable and complementary approach to modeling military working dog core temperature.
• The K9-TempML model shows potential for accurate temperature change estimation and feature importance analysis.
• Future work will focus on real-world data training and integration into practical tools for military applications.