A health index is not a diagnosis. It is a way of reading records you already have—weight, vaccinations, parasite prevention, medications, ongoing conditions—as one picture instead of six.
One number is only useful when it comes with its reasoning. The score is the entry point; "what's helping" and "what to watch" are the part you act on.
Most pet owners are not short of health information. They are short of a way to hold it all at once.
The vaccination certificate is in an email. The weight is on a card from the last visit. The parasite prevention is a monthly reminder you sometimes miss. The arthritis diagnosis is in a discharge note from eighteen months ago. Each piece is legible on its own. Together they never quite resolve into an answer to the question you actually have: is my pet doing well right now, and what should I be paying attention to?
A health index—sometimes called a pet health score or wellness score—is an attempt to answer that. It reads the record you have already built and returns a single reading, usually on a 0-100 scale, with a plain band and the reasoning behind it.
This post covers what belongs in that reading, where AI does real work, and the specific ways a score like this can mislead you if you read it wrong.
Why the record gets fragmented in the first place
Australian households now keep an estimated 31.6 million pets across 7.7 million homes—about 73% of all households.1 Very few of those homes have a single place where a pet's health history lives.
That is not carelessness. It is structural. A typical adult dog's record is generated by at least four separate systems:
- The clinic's practice-management software, which holds consultations, vaccinations and lab results—and which you generally cannot query yourself
- The product you buy, which sets the parasite-prevention cadence
- The pharmacy or repeat script, which sets the medication cadence
- Your own memory, which holds everything nobody wrote down: the limp that lasted four days, the week they went off their food
None of these talks to the others. So the practical question—"how is my pet tracking?"—gets answered once a year, in a ten-minute consult, by the one person who has seen all of it.
A health index does not replace that consult. It gives you something to bring to it.
What actually goes into a pet health index
A credible index is not a mood reading. It is a structured pass over a defined set of health factors, each of which has an established evidence base in veterinary practice. The AAHA Canine Life Stage Guidelines set out ten such factors that should be managed at each life stage, including nutrition, parasite control, vaccination, dental health, behaviour and a baseline diagnostic profile.2 That list is a reasonable shape for what an index should look at.
| Factor | What it reads | Why it carries weight |
|---|---|---|
| Body condition | Body condition score (BCS), weight | WSAVA's 9-point scale puts only 4 and 5 in the ideal band; 6-7 is overweight, 8-9 obese3 |
| Weight trend | Direction and rate of change over time | A trend can be meaningful before any single value looks abnormal2 |
| Vaccination | Core vaccine dates against the recommended interval | WSAVA recommends triennial or less frequent revaccination with core products in adult animals4 |
| Parasite prevention | Product, cadence, gaps in the log | Prevention is only protective at the interval it is dosed |
| Dental | Last dental assessment, noted disease | Periodontal disease is the most common disease of adult dogs and cats5 |
| Chronic conditions | Diagnoses on file and how they are managed | A managed condition and an unmanaged one are not the same signal |
| Medications | Current drugs, doses, as-needed use | Active, appropriate management is a positive, not a negative |
| Life stage | Age against breed-typical expectations | The same finding means different things at 2 and at 12 |
Two things follow from that table.
First, most of these are preventive-care facts, not clinical findings. Whether a dog's C5 is current is a matter of record. Whether a dog has early kidney disease is not—it needs bloods and a vet. An index built on the profile can be confident about the first category and must stay quiet about the second.
Second, the factors interact. Body condition is the clearest illustration. A survey of dogs examined at Australian veterinary practices found that of 2,661 dogs, 33.5% were overweight and a further 7.6% were obese6—so a raised BCS is common enough that, flagged on its own, it is close to background noise. What changes its meaning is what sits beside it. A BCS of 7/9 in a four-year-old Kelpie is a weight problem. The same 7/9 in a twelve-year-old Labrador with diagnosed hip osteoarthritis is a weight problem that is actively loading a painful joint. The second is worth flagging harder. Reading factors in combination, rather than as a checklist, is the part that is genuinely hard to do by hand—and the part where AI earns its place.
A worked example
Take a fictional pet—call him Rex, a six-year-old Labrador. His profile holds:
- Weight: 38.4 kg, up 3.1 kg over the last 14 months
- Body condition: 7/9 at the most recent assessment
- Vaccination: C5, last given 11 months ago—current
- Parasite prevention: monthly product, logged in 11 of the last 12 months
- Chronic condition: mild hip osteoarthritis, diagnosed 2 years ago, described as stiff after long walks
- Medications: daily joint supplement, plus an anti-inflammatory as needed on bad days
- Dental: last assessment 2 years ago
Read as a list, that is ambiguous. Read as a picture, it resolves:
What's helping. Preventive care is genuinely strong—core vaccination current, parasite prevention at 11/12 months. The osteoarthritis is not just diagnosed, it is being actively managed with a daily supplement and appropriate as-needed analgesia. That is the behaviour of an engaged owner, and it should count in Rex's favour.
What to watch. The 3.1 kg gain and BCS of 7/9 are the thread worth pulling, specifically because of the hip osteoarthritis. Weight and joint disease compound. Neither factor alone would sit at the top of the list; together they do.
What's missing. Two years without a dental assessment is a gap in the record, not a finding. The honest output is "we can't see this," not a guess.
A reading in the high 70s with a Good band is a defensible summary of that. The number is not the useful part. The ordering is—it tells Rex's owner that the next conversation with their vet should be about weight and the hip, not about vaccinations.
Where AI genuinely helps
Strip away the marketing and there are five concrete jobs here that language models are actually well suited to:
- Normalising messy input. "NexGard Spectra monthly," "flea/tick/worm chew," and "monthly all-wormer" are the same fact written three ways. Turning free-text owner entries into a comparable field is unglamorous and exactly the sort of thing AI does well.
- Reading factors in combination. The weight-plus-arthritis interaction above is a reasoning step, not a lookup. A rules engine would need that pairing hard-coded; a model can surface it from the record.
- Detecting trends across entries. Comparing a weight logged today against five logged over two years, and describing the direction, is a different task from checking whether today's value is in range. AAHA's guidance makes the point that trending an individual animal's own baseline can surface change while every value is still technically normal.2
- Naming what's missing. The genuinely valuable behaviour is refusing to score confidently on a thin profile and asking for the specific field that would change the answer.
- Explaining the reasoning in plain language. A score with no explanation is a horoscope. The explanation is the product.
Grouped output beats a flat list. Each line under "what to watch" is a prompt for the next conversation with your vet, not a finding the app has made on its own.
Note what is not on that list: deciding whether your pet is sick. Every item above is a records task. None is a clinical judgement.
Honest limitations: five ways a health score can mislead you
This is the section that matters most, so read it before you read your pet's number.
1. It only knows what you told it. An index computed from a profile is computed from a profile. A pet with a serious undiagnosed condition and a tidy vaccination record can score well. A healthy pet with a half-empty profile can score poorly. The score measures the record, and the record is not the animal.
2. A good score is not a clean bill of health. Nothing in a profile detects a heart murmur, an abdominal mass, dental pain under the gumline or early kidney disease. Those need a physical examination and, often, bloods. A high score should never be the reason you skip a check-up.
3. AI can state wrong things confidently. NIST's generative-AI risk profile names this failure mode "confabulation"—fluent, plausible output that is simply incorrect—and specifically warns that inaccurate summaries of health information can mislead people making consequential decisions.7 If a summary references a vaccination or a diagnosis you do not recognise, trust your record over the summary.
4. Clinical decisions are not delegable. The RCVS is explicit that clinical decision-making must not be wholly delegated to AI and that generated output requires critical review and human responsibility.8 That applies to owners too. A score is an input to a conversation, not a substitute for one.
5. Small movements are noise. A score that moves from 78 to 81 because you logged a worming dose has not told you your pet got healthier. Watch the band and the direction over months, not the digits week to week.
When to ignore the index entirely
Do not open an app when your pet is:
- Having difficulty breathing, collapsed, or seizing repeatedly
- Bleeding heavily, or showing a distended, painful abdomen with unproductive retching
- Straining to urinate without producing urine—particularly a male cat
- Deteriorating rapidly, or in obvious severe pain
Call your vet or the nearest emergency clinic. A profile-based score has nothing useful to contribute to an emergency, and the minutes matter.
What this means for pet owners
The realistic value of a health index is narrow and worth having.
It will not tell you what is wrong with your pet. It will tell you what your own records already imply and you have not had the chance to notice: that the weight has crept up 3 kg over a year, that the parasite prevention has a two-month hole in it, that you have not recorded a dental assessment since 2024.
That is a shortlist for the next vet visit. Owners who arrive with "I've noticed he's gained weight and I'm worried about the hip" get a more useful ten minutes than owners who arrive with "he seems fine, I think?"
Read it that way—as preparation, not as a verdict—and the number does real work.
Curious what your own records already say? Complete your pet's profile on PetCare AI and check their Health Index—you'll get a 0-100 reading, the reasoning grouped into what's helping and what to watch, and a clear list of what's still missing from the picture.
Sources
Footnotes
-
Animal Medicines Australia, "Pets in Australia: A national survey of pets and people 2025," September 2025. ↩
-
American Animal Hospital Association, "2019 AAHA Canine Life Stage Guidelines," Journal of the American Animal Hospital Association, 2019; see also AAHA, "Baseline Data." ↩ ↩2 ↩3
-
World Small Animal Veterinary Association Global Nutrition Committee, "Body Condition Score—Dogs," 2025. ↩
-
Squires RA et al., "2024 guidelines for the vaccination of dogs and cats," compiled by the WSAVA Vaccination Guidelines Group, Journal of Small Animal Practice, 2024. ↩
-
Merck Veterinary Manual, "Periodontal Disease in Small Animals." ↩
-
McGreevy PD et al., "Prevalence of obesity in dogs examined by Australian veterinary practices and the risk factors involved," Veterinary Record, 2005;156(22):695-702. ↩
-
National Institute of Standards and Technology, "Artificial Intelligence Risk Management Framework: Generative Artificial Intelligence Profile," NIST AI 600-1, July 2024. ↩
-
Royal College of Veterinary Surgeons, "Using artificial intelligence (AI) in practice—advice for the profession." ↩
