Objective Video-Based Assessment of ADHD-Like Canine Behavior Using Machine Learning.

Area of Science:

• Veterinary Behavior Science
• Machine Learning in Animal Health
• Canine Behavioral Diagnostics

Background:

• Canine ADHD-like behavior negatively impacts dog well-being and owner quality of life.
• Current diagnostic methods rely on subjective owner reports and assessment scales.
• Objective diagnostic tools are needed for timely and effective intervention.

Purpose of the Study:

• To develop and validate an objective, automated method for assessing canine ADHD-like behavior.
• To utilize machine learning for analyzing video data from a clinical setting.
• To establish a quantitative score (H-score) for behavioral assessment.

Main Methods:

• Trained a machine learning classifier on video data of dogs with and without ADHD-like behavior.
• Achieved 81% accuracy in differentiating between clinical and control groups.
• Developed an H-score based on classifier output to quantify behavioral severity.

Main Results:

• The machine learning model successfully differentiated between dogs with and without ADHD-like behavior.
• Preliminary clinical evaluation showed a reduction in H-score for 8 out of 11 dogs undergoing treatment.
• Expert feedback suggests potential clinical applications for the H-score.

Conclusions:

• Automated video analysis offers an objective approach to diagnosing canine ADHD-like behavior.
• The developed H-score shows potential as a reliable measure for treatment monitoring.
• Further research and validation are warranted for clinical implementation.