Describe only what the image can support.
PetCheck records visible form, color, coating, contents, distribution, and attention areas. A photograph can support an appearance record; it cannot identify a disease or internal cause.
EVIDENCE-AWARE · NON-DIAGNOSTIC
Research should make uncertainty clearer — not make an app sound like a veterinarian.
PetCheck uses peer-reviewed veterinary research to improve how a stool photo is described, how owner context is organized, and how changes are followed over time. It does not use those papers to claim a diagnosis, cause, prognosis, or treatment.
Original evidence catalog date: 2026-09-01WHAT THE EVIDENCE CAN SUPPORTMeasurement · context · repeated observation · safer handoff
WHAT PETCHECK WILL NOT CLAIMDisease · cause · treatment · replacement for veterinary care
Who made the Learn guides?
PetCheck editorial prepares the Learn guides with AI-assisted research, drafting, and visual production. The guides explain observations and help owners prepare questions for their veterinarian. They have not been independently veterinarian-reviewed or validated as diagnostic material.
Each guide names the sources used for its specific safety or observation points. University and public-health guidance supports those points, not PetCheck accuracy. A research abstract is identified as an abstract; it is not presented as a full-paper review.
Article dates describe editorial changes to that article. An edited date is not a veterinarian’s sign-off or a claim that every cited source was rechecked on that date. When a source-check date is given, it applies only to the sources identified with it.
Editorial photographs and AI-created scenes illustrate the topic; they are not patient cases or evidence that a pictured animal has a condition. Image captions identify their role. Comparison drawings are instructional models, not a calibrated scoring chart. Fictional notes and brief previews are labeled and are never presented as customer outcomes or saved as real pet records.
Practice questions help explain the reading. Their answers do not produce a diagnosis, treatment plan, or clinical score. Educational examples are kept separate from observations you choose to record in PetCheck.
To flag an unclear statement, source, or image caption, contact PetCheck with the article link and the passage you mean. Content feedback is not a route for urgent veterinary advice.
Return to the visual guides →Four layers, one conservative next step.
The photo is deliberately not allowed to dominate the decision. Each layer has a different job and a visible limit.
Let the whole pet outweigh a single color.
Frequency, appetite, energy, vomiting, water intake, pain or straining, life stage, recent food changes, and existing care can move the safety boundary.
Separate one moment from a repeating pattern.
A dated sequence can show whether visible appearance and owner-reported context are improving, stable, repeating, or worsening. The sequence still does not diagnose a cause.
Stop before diagnosis or treatment.
PetCheck provides one conservative next step: keep, observe, review food fit after another record, or contact a veterinarian first. Treatment remains outside the product.
HOW TO READ A PETCHECK RESULT
Confidence describes the image reading, not the pet's health.
A high-confidence photo can still belong to an unwell pet. A low-confidence photo may simply have glare, blur, an incomplete frame, a difficult background, or visually similar material. Owner confirmation and whole-pet context remain separate layers.
Visible features are easier to describe. Medical certainty does not increase.
The automatic description should be checked before it informs the next step.
PetCheck preserves uncertainty and pauses product matching when needed.
What each paper supports — and what it does not prove.
Scope limits are part of the evidence. Puppy studies are not presented as feline validation. Computer-vision research is not presented as PetCheck accuracy. Clinical indices are not copied into a consumer diagnostic score.
Photo method validation2026
Validation of photographic fecal scoring in puppies
Chauvel C, Bendahmane I, Vellard E, et al.
- Population and design
- Puppies; 40 fresh samples and 98 stool photographs. Method-comparison and observer-agreement study.
- Plain-language takeaway
- A clear photo can be useful for recording consistency in puppies, but the photo is still only an appearance record.
- Study snapshot
- 40 fresh puppy stool samples were compared in situ and by photograph; 98 photographs were scored by four observers. Weighted kappa was 0.93 for method agreement and ICC was 0.83 for inter-observer agreement.
- Supports
- A puppy fecal-consistency scale can be applied to photographs with good agreement under the study conditions.
- Does not prove
- It does not validate disease diagnosis, cause detection, treatment selection, adult dogs, cats, or PetCheck as a medical device.
Puppy scale validation2012
Validation of a fecal scoring scale in puppies during the weaning period
Grellet A, Feugier A, Chastant-Maillard S, et al.
- Population and design
- Weaning puppies. Fecal-scale validation study.
- Plain-language takeaway
- A stool score is a measurement aid, not a diagnosis; the pet's age and context still matter.
- Study snapshot
- The study used a 13-point scale during weaning and showed that puppy age and breed size affected how fecal quality should be interpreted.
- Supports
- Fecal quality can be described with a structured scale, and age and breed size can affect interpretation in puppies.
- Does not prove
- It does not establish a universal normal score for every dog, and it does not diagnose the cause of an abnormal stool.
Observer reliability2021
Consistency of faecal scoring using two canine faecal scoring systems
Cavett CL, Tonero M, Marks SL, et al.
- Population and design
- Dogs; veterinarians and members of the public. Agreement study using two fecal-scoring systems.
- Plain-language takeaway
- Two people may describe the same stool differently. PetCheck therefore shows uncertainty and asks the owner to confirm visible signals.
- Study snapshot
- Naturally passed stools from 126 dogs were scored by three veterinarians and 126 members of the public. Agreement varied, and lay-veterinarian agreement was only fair.
- Supports
- Different observers can assign different fecal scores, and experience improves agreement without eliminating variation.
- Does not prove
- It does not make an individual photo objectively correct, and it does not turn a fecal score into a diagnosis.
Computer-vision feasibility2023
A Fine-Grained Image Classification Approach for Dog Feces Using MC-SCMNet under Complex Backgrounds
Liang J, Cai W, Xu Z, et al.
- Population and design
- Dog-feces image dataset under complex backgrounds. Machine-learning image-classification study.
- Plain-language takeaway
- Image AI can sort visible patterns, but dirt, lighting, weathering, framing, and look-alike features can still mislead it.
- Study snapshot
- On the authors' self-built DFML dataset, the research model reported 88.27% average identification accuracy and an F1 value of 88.91%; those figures do not describe PetCheck.
- Supports
- Fine-grained automated classification of dog-feces images is technically feasible, while background and small visual differences remain challenging.
- Does not prove
- Its reported model accuracy is not PetCheck accuracy and is not clinical diagnostic validation.
Multidomain clinical context2020
Effect of amoxicillin-clavulanic acid on clinical scores, intestinal microbiome, and amoxicillin-resistant Escherichia coli in dogs with uncomplicated acute diarrhea
Werner M, Suchodolski JS, Straubinger RK, et al.
- Population and design
- Dogs with uncomplicated acute diarrhea. Randomized, placebo-controlled clinical study.
- Plain-language takeaway
- The whole pet matters. Energy, appetite, vomiting, water intake, and frequency can change the safety boundary even when a photo looks mild.
- Study snapshot
- Sixteen client-owned dogs were followed with a canine acute diarrhea severity index that combined activity, appetite, vomiting, fecal consistency, and defecation frequency.
- Supports
- Veterinary research tracks activity, appetite, vomiting, fecal consistency, and defecation frequency together rather than relying on stool appearance alone.
- Does not prove
- PetCheck does not reproduce the study's clinical score, assess dehydration, or recommend medication.
Evidence-based veterinary guideline2024
European Network for Optimization of Veterinary Antimicrobial Therapy (ENOVAT) guidelines for antimicrobial use in canine acute diarrhoea
Jessen LR, Werner M, Singleton D, et al.
- Population and design
- Dogs with acute diarrhea; veterinary decision-making. Multidisciplinary evidence-based guideline.
- Plain-language takeaway
- PetCheck stops at organized observation. Treatment choices remain with a veterinarian who can examine the dog and apply clinical guidance.
- Study snapshot
- A multidisciplinary panel translated a systematic evidence review into veterinary recommendations for antimicrobial and probiotic use in canine acute diarrhea.
- Supports
- Treatment decisions for canine acute diarrhea belong in an evidence-based veterinary assessment, not an image-only consumer tool.
- Does not prove
- PetCheck does not use the guideline to prescribe, withhold, or compare antimicrobials, probiotics, fluids, or any treatment.
Owner-recorded feline pattern2022
Survey of defecation habits in apparently healthy and chronic kidney disease cats
Jones SE, Quimby JM, Summers SC, et al.
- Population and design
- Apparently healthy cats and cats with chronic kidney disease. Owner survey plus repeated daily observations.
- Plain-language takeaway
- For cats, a short run of actual observations is more useful than relying on memory about what is 'usual.'
- Study snapshot
- The study analyzed survey data from owners of apparently healthy cats and cats with chronic kidney disease, including repeated daily observations rather than recall alone.
- Supports
- Repeated owner observations can characterize defecation frequency and fecal scores more directly than a one-time recollection.
- Does not prove
- Frequency alone does not identify disease, and findings from cats with chronic kidney disease should not be generalized to every cat.
Feline multidomain context2010
A clinical index for disease activity in cats with chronic enteropathy
Jergens AE, Crandell JM, Evans R, et al.
- Population and design
- Cats with chronic enteropathy in veterinary care. Clinical-index development and validation study.
- Plain-language takeaway
- A cat's stool is one clue among many. PetCheck uses simple owner-observed context only to set a safer next step.
- Study snapshot
- The index was developed across 82 cats: 59 retrospective cases and 23 prospectively assessed cats. It combined clinical, endoscopic, and laboratory variables unavailable to a home photo tool.
- Supports
- Feline gastrointestinal disease activity is assessed from multiple clinical and veterinary variables, not stool appearance in isolation.
- Does not prove
- PetCheck cannot calculate this veterinary index and does not infer chronic enteropathy from a home photo or questionnaire.
Terms that should never be left unexplained.
Technical language is useful only when its meaning and limits are visible.
Fecal consistency
How formed, soft, loose, or watery the stool appears.Consistency is a visible description. It does not reveal the disease, organism, food, or internal process that caused the appearance.
Whole-pet or systemic signs
Changes beyond the stool itself, such as energy, appetite, vomiting, or ability to keep water down.These observations can move a case across PetCheck's safety boundary. They are not a substitute for a physical examination or hydration assessment.
Image confidence
How reliable the visible-signal reading appears to be.It describes the image reading, not the chance that a pet is healthy. Lighting, framing, background, blur, and look-alike material can reduce confidence.
Longitudinal pattern
A pattern built from observations collected across time.Two or more dated records can show whether appearance and context are improving, stable, repeating, or worsening. The pattern still does not diagnose a cause.
Validation
Testing whether a method performs as intended under stated study conditions.Validation is limited to the population, procedure, and outcome that were studied. It should not be stretched into claims about other species, diseases, or products.
Inter-observer agreement
How often different people give the same or similar rating.Lower agreement means human review can vary. PetCheck therefore keeps uncertainty visible and lets the owner correct the automatic description.
Weighted kappa
A chance-adjusted agreement statistic for ordered ratings.It gives partial credit when two ratings are close rather than identical. Values closer to 1 indicate stronger agreement, but they do not prove clinical diagnosis.
Intraclass correlation coefficient (ICC)
A statistic used to describe reliability across multiple raters or measurements.Higher values indicate more consistent ratings in the studied setting. The statistic applies to the measurement method, not to a pet's health certainty.
Clinical index
A structured set of variables used by clinicians or researchers to summarize severity.A clinical index may include examination or laboratory data that PetCheck cannot obtain. PetCheck does not calculate or imitate a veterinary clinical score.
Non-diagnostic
Describes and organizes observations without identifying a disease or cause.PetCheck can help prepare a clearer record and a safer next step. It cannot rule a condition in or out, prescribe treatment, or replace veterinary care.
ONE PHOTO · CLEAR LIMITS · ONE NEXT STEP
Start a stool checkUse the evidence method on a real observation.
No account is required for the first scan. PetCheck will show what the image supports, what it cannot tell, and what to record next.