FOR IMMEDIATE RELEASE
Vetology Expands Public AI Validation Dashboard to 11 Metrics Per Condition Classifier, Commits to Ongoing Model Retraining
Vetology publishes full statistical profiles for 89+ classifiers, reinforcing that building AI is only half the job.
March 31, 2026 – SAN DIEGO – Vetology, a provider of AI-generated radiology screening reports and board-certified veterinary teleradiology, today expanded its publicly available AI performance dashboard from four metrics per classifier to eleven. The update covers 89+ validated classifiers across canine and feline thoracic, abdominal, and musculoskeletal imaging.
The expanded dashboard now reports sensitivity, specificity, positive predictive value, negative predictive value, AUC, F1 score, accuracy, prevalence, confidence intervals, and Radiologist Agreement Rate for each condition. The data is available on our website.
The update also reflects Vetology’s commitment to maintaining its existing classifiers alongside building new ones. Of the 89+ classifiers currently published, 31 are retrained models that were originally released, then revalidated against updated board-certified radiologist consensus data. The oldest classifiers in the current dashboard date to August 2024; all have been revalidated with confusion matrices generated as recently as February 2026.
“AI is changing fast, and we are working just as hard to keep pace. We put the same rigor into maintaining our older models as we do into building new ones. Publishing the data for all of them, new and retrained, is how we honor our commitment to our veterinary partners and patients.”
Eric Goldman, President, Vetology
New classifiers added in this update include Obscuring Pleural Effusion, Esophageal Enlargement, Intervertebral Disc Disease (Thoracic), Small Intestine Enlargement (Feline), Colon Diffuse Distension (Feline), and a consolidated Heart Failure classifier for canine imaging. The Heart Failure classifier reports 89.5% sensitivity and 92.1% specificity; the Obscuring Pleural Effusion classifier, designed to flag cases where fluid volume may limit diagnostic interpretation, reports 87.2% sensitivity and 96.7% specificity.
“We’re improving our classifiers every month, and every update is revalidated against fresh consensus reads from board-certified radiologists – not the same training set warmed over. That’s why we publish eleven metrics per classifier instead of the one or two you’ll see from other vendors. Sensitivity by itself doesn’t tell a clinician whether to trust a result. PPV, confidence intervals, specificity – that’s what lets a veterinarian decide how much weight to put on what the model is telling them. We think that level of transparency should be the baseline for veterinary imaging AI. As far as we can tell, nobody else is publishing it.”
Cory Clemmons, CTO, Vetology
Vetology’s validation data is built on a foundation of 300,000 multi-image patient cases, with classifier performance validated against board-certified veterinary radiologist consensus. The company publishes these metrics as part of its commitment to transparency in an industry where, according to a 2026 Frontiers in Veterinary Science audit, 63.3% of commercial veterinary AI vendors do not disclose validation data publicly.
ABOUT VETOLOGY
Vetology provides AI-generated radiology screening reports and on-demand teleradiology consultations from board-certified veterinary radiologists, cardiologists and a dentist, giving veterinary practices both speed and specialist depth in a single platform. The Vetology AI screening system covers a growing list of conditions across canine and feline thoracic, abdominal, and musculoskeletal imaging. Screening results are designed to fit naturally into existing clinic workflows, so veterinary teams can move from image to informed decision without adding steps to their day. Vetology was founded on the belief that humans and AI are better together.
Learn more at vetology.net.