Portable Electronic Olfactometer for Non-Invasive Screening of Canine Ehrlichiosis: A Proof-of-Concept Study Using...

Area of Science:

• Veterinary Medicine
• Analytical Chemistry
• Machine Learning

Background:

• Canine ehrlichiosis, caused by Ehrlichia canis, poses a significant challenge, especially in resource-limited areas with restricted diagnostic access.
• Current diagnostics for canine ehrlichiosis are often laboratory-based, limiting their accessibility and speed in certain settings.

Purpose of the Study:

• To evaluate the feasibility of a portable electronic olfactometer for non-invasive screening of canine ehrlichiosis.
• To analyze volatile organic compounds (VOCs) in breath, saliva, and hair samples for disease detection.
• To assess the performance of machine learning classifiers in identifying infected dogs based on sensor data.

Main Methods:

• Acquired signals using an array of eight metal-oxide (MOX) gas sensors from breath, saliva, and hair samples of 38 dogs (19 infected, 19 controls).
• Applied principal component analysis (PCA) for dimensionality reduction on sensor data.
• Utilized supervised machine learning classifiers (AdaBoost, SVM, k-NN, RF) to analyze features and classify samples.

Main Results:

• Support Vector Machine (SVM) models demonstrated the strongest performance, particularly with saliva samples, achieving 94.7% accuracy, sensitivity, and precision (AUC = 0.964).
• Breath and hair samples exhibited lower discriminative power compared to saliva.
• The study involved 114 samples across three biological matrices from PCR-confirmed cases and controls.

Conclusions:

• Electronic olfactometry shows potential as a complementary, low-cost, non-invasive screening tool for canine ehrlichiosis.
• Saliva analysis using SVM appears most promising for this screening approach.
• Further research with larger sample sizes is warranted to validate these preliminary findings.