Can a Smart Litter Box Detect Kidney Disease? What the New Study Actua

Search for an automatic litter box in 2026 and you will meet this claim within about thirty seconds: a smart litter box can spot kidney disease before you can.

It appears on manufacturer blogs, on retailer product pages, in comparison articles that are really advertisements. It is usually attached to a sentence about a real scientific study. And there is a real study — a new one, published in May 2026, in a peer-reviewed journal.

We went and read it.

The study is real, the findings are statistically solid, and almost every article quoting it is describing something the study did not do.

Before you read this: we sell one of these

We sell an automatic litter box. So we have an obvious interest in you believing that automatic litter boxes are medically useful.

This article argues the opposite of that interest. It ends with three things we cannot tell you about our own machine, and with a recommendation that has nothing to do with buying anything from us.

Read it with that in mind.

First, why anyone cares

Chronic kidney disease is not a rare thing that happens to unlucky cats. It is the thing that happens to old cats.

The ISFM Consensus Guidelines on the Diagnosis and Management of Feline Chronic Kidney Disease (Sparkes et al., Journal of Feline Medicine and Surgery, 2016) put it in numbers:

About 4% of all cats seen in general practice — the seventh most common specific diagnosis
At least 30–40% of cats over ten years old
Renal disease was the most common cause of death in cats aged five and over, accounting for more than 13% of deaths, at a median age of 15

So the interest in early detection is not manufactured. If your cat lives long enough, this is statistically the most likely thing to be wrong with it. And cats are famously good at hiding illness until it is advanced.

A device that sits in the room the cat visits several times a day, and weighs it every time, is genuinely an interesting place to look for early signals. That part of the pitch is not silly.

What the study actually did

The paper is Langenfeld-McCoy et al., "Enhancing Detection of Feline Chronic Kidney Disease Through Smart Litter Box Monitoring," Animals, 2026, volume 16, issue 9, article 1319.

It compared cats with confirmed CKD against healthy cats, using data a smart litter box monitor collected between January 2023 and September 2025. The monitor recorded visit frequency, how long the cat took to eliminate, the weight of urine produced, and how long and how vigorously the cat covered afterwards.

Here is what it found.

Measured over a week	Cats with CKD	Cats with no known condition	p
Urination events	4.6	2.3	<0.001
Time spent voiding	25.2 s	17.8 s	0.015
Time spent covering afterwards	14.0 s	22.9 s	0.001
Weight of urine produced	1,007.4 g	469.8 g	0.015

Those differences are large and they are statistically significant. A cat with kidney disease produced roughly twice as much urine, twice as often, spent longer doing it, and spent less time covering it up. The model built on those signals scored a weighted F1 of 92.7% on the training data and 89.9% on a separate validation set.

That is a real result and we are not going to pretend otherwise. If you had asked us beforehand whether litter box behaviour carries a detectable kidney signal, we would have guessed yes but had nothing to point at. Now there is something to point at.

Now the four things the marketing leaves out

1. It was retrospective. It does not show early detection.

This is the big one, and it is the difference between what the paper demonstrates and what the headlines say.

The study compared cats that had already been diagnosed with cats that had not. It did not follow healthy cats forward through time and catch the moment they started to change. There is no before-and-after. There is no timeline.

So the paper cannot tell you how many weeks or months ahead of a vet these signals appear — because it did not measure that. The paper's own limitations section says the design was retrospective and had no longitudinal follow-up.

"Cats with kidney disease pee more than cats without it" and "a litter box will warn you before your vet does" are two completely different claims. The study supports the first. Every article we found was selling the second.

2. Seven of the nine authors work for the company that sells the device

The author affiliations are Nestlé Purina Research and Ohio State University. Seven authors are employed by Nestlé Purina Research. An eighth serves on a Purina advisory board. All of this is disclosed properly in the paper's conflict of interest statement — the researchers did the honest thing.

The funding line reads "This research received no external funding." Which, read plainly, means Purina paid for it internally.

Purina sells a smart litter box monitor.

This does not make the study wrong. Peer review happened, the conflict was declared, the statistics are what they are. Industry funds a great deal of legitimate veterinary science. But a reader deciding whether to spend several hundred dollars deserves to know who paid for the evidence, and not one of the articles citing this paper mentioned it.

3. The sample is small, and some sick cats looked healthy

The training set was 13 cats with CKD against 72 healthy cats. The validation set was 7 against 44. Data per cat ranged from two weeks to two years, with a median of six weeks.

Thirteen sick cats is not many. The authors say so themselves, and list the limitation plainly: small sample size, limited age range among the CKD cats, and health status for the at-home cats confirmed only by what the owner reported.

And then this, which we think is the single most useful sentence in the paper for anyone actually considering a purchase: some of the cats with CKD had behavioural profiles indistinguishable from healthy cats.

4. It was deliberately tuned to miss sick cats rather than alarm healthy ones

The model was built to prioritise precision over sensitivity. In plain language: when it says a cat is sick, it is usually right — and the price of that is false negatives.

It is a reasonable engineering choice. Nobody wants a machine that panics every week. But it has a consequence that should be printed on the box:

The machine saying nothing is not the same as your cat being fine.

The detail that tells you the most

Here is the part we did not expect.

The device behind this research is Purina's Petivity Smart Litter Box Monitor, which grew out of a larger project — reported in Scientific American — covering 191 cats and more than 300,000 recorded litter box visits since 2018.

And Purina's own position on what it sells is that the monitor is "not intended to diagnose, treat, cure, or prevent any disease."

That is the manufacturer, who funded the research, describing the limits of their own product, in language considerably more careful than the retailers quoting their study.

When the company that paid for the evidence is more cautious about the claim than the shops repeating it, that tells you where the exaggeration is entering.

A Singapore-shaped problem nobody mentions

Two operating constraints from that same reporting are worth knowing before anyone here buys a monitor for health reasons.

The first: the system distinguishes individual cats by weight and behaviour, and it struggles when cats are similar in weight with similar habits. The recommendation is five cats or fewer per system.

The second is the one that matters locally. Vibration and movement from nearby appliances — robot vacuums, washing machines — can confuse the readings.

Now go and look at where the litter box lives in most flats here. It is in the service yard. It is in the service yard because that is where the drain and the floor tiles are. And as we noted when we surveyed this category, service yards in older HDB flats often have exactly one power point — and the washing machine is in it.

So the standard Singapore installation is: a weight-sensing health monitor, on a tiled floor, next to a spinning washing machine, sharing its socket. Nobody selling one here has mentioned this to you.

What we cannot tell you about our own machine

This is the part where we would normally explain that our box does all of this. We are not going to, because we do not know that it does, and we suspect it does not.

Our machine has a gravity sensor among its six detection systems and it has an app. That is not the same thing as what this research used. The study measured seconds spent voiding, seconds spent covering, grams of urine and per-visit identity. That needs purpose-built sensing and a trained model behind it, not a load cell and a notification.

So, plainly, three things we do not know about our own product:

Does the app record weight per visit, or only detect that a cat is present?
Does it log visit duration and frequency over time, and can that history be exported?
Is there any retained record at all after the notification disappears?

We have asked our manufacturer. When we have the answers they go on the product page, whichever way they fall. If the answer is that it records nothing useful, we will say that, because a customer who buys our machine expecting health tracking and discovers a rake on a timer has been misled by us.

This is now the third specification we have had to publish as a blank rather than a number — alongside the minimum detection weight, which we still do not have, and the bulk density of our own litter, which we are measuring ourselves. We would rather the list of blanks be public and shrinking than private and comfortable.

So what should you actually do

If you are worried about your cat's kidneys, the thing that answers the question is a blood and urine test at a vet. It is not a litter box, at any price, from anyone, including us. The ISFM guidelines exist precisely because this disease is diagnosed and staged on laboratory values.

What a litter box can genuinely give you is something narrower and still useful: a record. Most owners cannot say how many times their cat urinated last Tuesday, or what it weighed six weeks ago. A machine that keeps that log turns "he seems a bit off lately" into something you can show someone.

That is a real benefit and it is worth paying something for. It is not a diagnosis and it does not replace an appointment.

And if the number the study leaned on hardest is the one you care about — twice the urine, twice as often — note that you do not strictly need an algorithm to notice a cat that has started producing double the wet litter. You need to be paying attention, and to have a rough idea of what normal looked like. The machine helps with the second part.

What would change our mind

We would take the health-monitoring pitch seriously — including from ourselves — when there is a study that does the thing this one did not: follow healthy cats forward, catch the ones who develop CKD, and report how far ahead of clinical diagnosis the signal appeared. Prospective, not retrospective. With a timeline in weeks.

That study is a reasonable thing to expect within a few years, because the data is being collected right now in ten thousand homes. When it is published, we will read it and write this article again.

Until then, the accurate sentence is: cats with kidney disease use the litter box differently, and a machine can measure the difference. Everything beyond that is currently a sales pitch wearing a citation.