Machine Learning Techniques for Canine Myxomatous Mitral Valve Disease Classification: Integrating Anamnesis, Quality of Life Survey, and Physical Examination

Engel-Manchado, J., Montoya-Alonso, J. A., Doménech, L., Monge-Utrilla, O., Reina-Doreste, Y., Matos, J. I., Caro-Vadillo, A., García-Guasch, L., & Redondo, J. I. (2024). Machine Learning Techniques for Canine Myxomatous Mitral Valve Disease Classification: Integrating Anamnesis, Quality of Life Survey, and Physical Examination. Veterinary Sciences, 11(3), 118. https://doi.org/10.3390/vetsci11030118.

This study explored the use of machine learning to classify canine myxomatous mitral valve disease (MMVD) using structured anamnesis, quality of life surveys, and physical examinations, rather than echocardiography. A sample of 1011 dogs from 23 hospitals was analyzed using both a classification tree and random forest model. The complex model had high accuracy for classifying healthy dogs (96.9%), but struggled with differentiating between stages B1 and B2 (49.8% and 62.2% accuracy, respectively) and advanced stages of the disease. A simplified model grouping B1 and B2 into stage B and C and D into stage CD improved accuracy, suggesting potential clinical utility despite limitations.