Research library
Selected to balance seminal findings, disease diversity, stronger blinded designs, negative or cautionary findings, and the most important recent developments. Links go directly to PubMed.
AllCancerInfectiousNeurologyMetabolic2026CancerMulti-cancer
In a 1,502-participant test cohort, a Bayesian fusion system combining multiple dogs’ indications reported 90.8% sensitivity, 91.3% specificity and AUC 0.962; early-stage sensitivity was 90.6%. The authors call for prospective evaluation in true screening populations.
Major 2026 updatePubMed · PMID 42024827 ↗2024NeurologyParkinson's disease
Twenty-three companion dogs were trained on sebum samples. Group performance averaged 89% sensitivity and 87% specificity; performance on novel samples averaged 86% sensitivity and 89% specificity.
Controlled studyPubMed · PMID 39352420 ↗2022CancerLung cancer
Two dogs were tested on serum or breath samples in unblinded, single-blind and double-blind settings. Double-blind accuracy varied by dog, illustrating both promise and reproducibility challenges.
Double-blindPubMed · PMID 34998991 ↗2021InfectiousCOVID-19
In a randomized double-blind study, dogs transferred scent recognition across saliva, urine and sweat, with reported sensitivity/specificity varying by specimen type.
Randomized double-blindPubMed · PMID 34315418 ↗2021InfectiousCOVID-19
Dogs distinguished SARS-CoV-2 samples from other respiratory viral infections with specificities above 90% across test scenarios, while sensitivities were lower and more variable.
Specificity challengePubMed · PMID 34869443 ↗2021CancerLung cancer
A double-blind study reported 97.6% detection of lung-cancer cases when breath and urine were combined; performance was lower when either specimen was used alone.
Double-blindPubMed · PMID 34388977 ↗2019CancerLung cancer
Three trained beagles distinguished lung-cancer serum from healthy controls with reported sensitivity of 96.7% and specificity of 97.5% in this double-blind study.
Double-blindPubMed · PMID 31206136 ↗2019NeurologyEpileptic seizures
Dogs were able to discriminate seizure-associated human odor samples in a controlled study, supporting the existence of an olfactory signature associated with seizures.
Proof of conceptPubMed · PMID 30923326 ↗2017CancerCervical cancer
A trained beagle distinguished cervical-cancer-related samples from controls in a double-blind protocol with reported sensitivity and specificity above 90%.
Double-blindPubMed · PMID 28122528 ↗2016InfectiousBacteriuria / UTI
In a double-blind validation study using 687 individuals, dogs detected bacteriuria caused by several common organisms with sensitivity at or near 100% and specificity above 90%.
Large validationPubMed · PMID 27186578 ↗2016CancerMelanoma
A dog selected melanoma samples above chance in 20 double-blind tests, supporting a melanoma-associated volatile organic compound signal while showing only modest absolute performance.
Proof of principlePubMed · PMID 27454583 ↗2015CancerProstate cancer
Two highly trained German Shepherds tested 362 prostate-cancer patients and 540 controls, with reported sensitivities of 100% and 98.6% and specificities of 98.7% and 97.6%.
Large diagnostic studyPubMed · PMID 25264338 ↗2015MetabolicHypoglycemia
Trained diabetes alert dogs identified hypoglycemia samples significantly above chance in vitro; sensitivity varied substantially between dogs while specificity was high.
Service-dog evidencePubMed · PMID 26440208 ↗2012InfectiousC. difficile infection
A trained beagle detected C. difficile in stool samples with 100% sensitivity and specificity in the test set and identified infected hospital patients with 83% sensitivity and 98% specificity.
Hospital proof of principlePubMed · PMID 23241268 ↗2011CancerColorectal cancer
A Labrador retriever distinguished colorectal-cancer breath and stool samples with reported sensitivity of 0.91/0.97 and specificity of 0.99/0.99, including early-stage cases.
Landmark studyPubMed · PMID 21282130 ↗2011CancerProstate cancer
A Belgian Malinois tested biopsy-characterized urine samples in a double-blind procedure and achieved reported sensitivity and specificity of 91%.
Double-blindPubMed · PMID 20970246 ↗2011CancerLung cancer
Dogs identified lung cancer from breath with 71% sensitivity and 93% specificity in 220 volunteers, including COPD controls; results suggested a robust volatile signature but imperfect performance.
Clinical challengePubMed · PMID 21852337 ↗2010CancerOvarian cancer
Two trained dogs detected an ovarian-cancer-associated odor in tissue and plasma, with high reported sensitivity and specificity in this early proof-of-concept work.
Early blood studyPubMed · PMID 21106067 ↗2006CancerBreast & lung cancer
Five household dogs were trained to distinguish breath samples. In double-blind testing on new donors, reported sensitivity/specificity were 0.99/0.99 for lung cancer and 0.88/0.98 for breast cancer.
Seminal breath studyPubMed · PMID 16484712 ↗2004CancerBladder cancer
Six dogs selected bladder-cancer urine above chance (41% versus 14% expected by chance), a landmark demonstration that tumor-related volatile compounds could be present in urine.
Landmark proof of principlePubMed · PMID 15388612 ↗Evidence standards
A 2022 systematic review of remote medical scent detection found 62 cancer and infectious-disease studies, but only a small number used double-blind, screening-like designs. Across the field, performance varies widely and bias/standardization remain major concerns. Cancer Dawgs should celebrate the dogs while being equally serious about experimental design.
Best practice
Testing on donors the dog has never encountered reduces memorization and leakage.
Best practice
Neither handler nor evaluator should know where the target is during formal testing.
Best practice
Healthy controls alone are not enough; realistic competing diagnoses matter.
Best practice
Sensitivity, specificity, predictive values and failures should be reported transparently.
Best practice
One-positive-in-five lineups can exaggerate performance compared with population screening.
Best practice
Independent teams, dogs and sample collections are needed before clinical claims.
The real promise may be bigger than a dog replacing a laboratory test. Dogs can help us discover which invisible chemical signals disease leaves behind.Cancer Dawgs · Canine Early Disease Detection for Happy Humans
Research summaries are educational and do not constitute medical advice or diagnostic services. Always consult qualified medical professionals and use validated clinical testing for healthcare decisions.
Curated research update: August 18, 2026. PubMed links are included for source verification.
We love our customers, so feel free to visit during normal business hours.
Summerfield Road & Secor Road, Lambertville, MI, USA
Open today | 09:00 am – 05:00 pm |
Sign up for our newsletter!
We use cookies to analyze website traffic and optimize your website experience. By accepting our use of cookies, your data will be aggregated with all other user data.