The Pet Food and Health app relies heavily on recent scientific work. This includes both technical achievements, such as data synchronisation and the technology underpinning the food blends, but obviously also a large body of work from the pet food and veterinary community. This page takes a look behind the scenes, or science, and gives a clear picture of how the Pet Food and Health app differs and excels from other food tools.
Food is not absorbed by pets in the same way that humans absorb food. While humans are omnivores, dogs are facultative carnivores whereas cats are obligatory carnivores. This affects how food is processed by their intestinal system. Understanding how pet food translates into metabolisable energy (i.e. the amount of energy a pet can obtain from eating these foods) is an active area of research. Many of these works continually try to improve on the status quo, often by taking more factors into account. For example, the simplest methods – and one proposed by the influential AAHA doi: 10.5326 – are variations of the formula used for humans:
kcal = 8.5 x fat + 3.5 x protein + 3.5 x carbsThis is also known as the Atwater equation. Many improvements on this formula exist doi: 10.17226 and many more have been considered and gotten in disuses over the decades. However, even when the formula is known, it's components can often be hard to obtain or calculate. For example, pet food labels often omit carbohydrates and moisture levels.
Work in the scientific community has more recently moved into (re-)incorporating gross energy as obtained from bomb calorimeters, and further differentiating between cats/dogs as well as dry/wet foods doi: 10.1371 doi: 10.3389 . Such approaches offer a novel and rich new way of more accurately assessing how our pets absorb the foods they get. And while it is always exciting to see improvements in this field, care should be taken to evaluate and critically assess these works. For these reasons, the app does not use these most recent approaches as the technical machinery to obtain the new formulas appears flawed, in part due to factors such as over-fitting and self-evaluation. Other studies confirmed these suspicions using empirical evaluation doi: 10.1371 , thus highlighting that the most accurate estimates remain those from previous work doi: 10.17226 . Therefore, the app uses established results for calculating the calorific content of pet food doi: 10.17226 while advancing on the simpler method explored in the AAHA guidelines and while keeping a keen eye on further progress in this area of research.
The current golden standards for pet nutrition are provided by the 2021 AAHA guidelines doi: 10.5326 and the 2021 FEDIAF nutritional guidelines. A limitation of these works is that they are designed for a very general and diverse audience. These provide the safe guidelines from which a better approach can be refined. Nevertheless, these works are highly relevant and they set a lot of the requirements that pet food manufacturers need to adhere to, such as micro-nutrient contents and limits or minima for macro-nutrients.
Both AAHA and FEDIAF give a solid starting point to determine the maintenance energy requirements for pets and dogs based on body weight. Yet, clearly, this is inadequate for detailed advise. Given four dogs of the same weight, one may be young and the other old, one may be lethargic whereas the other is a working dog. Clearly their energy requirements will be different.
The app firmly adopts some of the most recent research on pet dog maintenance energy requirements doi: 10.1371 and augments this with a rich multi-faceted approach to carefully establish the maintenance energy requirements, not only based on age and size, but also on their lifestyle factors doi: 10.1017 .
For cats in particular this approach can be notably refined based on recent research. The app starts from a strong basis doi: 10.1017 , but then refines further to take the feline body mass index (fBMI) doi: 10.1053 into account. The reason for this difference when dealing with cats is that the cat maintenance energy requirements show quite a large spread. This spread is greatly reduced when body mass is ignored in favour of lean body mass (i.e. disregarding body fat) doi: 10.1177 . This is obviously an additional calculations – and the owner needs to have a well-behaved cat to calculate the fBMI – so the app defaults to the more generic calculation when no fBMI is known, but relies on the more precise and recent values when the user takes the time to calculate their cat's fBMI.
Minima for proteins in pet food are provided by both AAHA and FEDIAF. These values change as a pet grows, are distinct for dogs and cats, and of course also vary with the size of the pet. However, these are not the be all, end all values. More recent work in Pratique Vétérinaire and others doi: 10.1371 establishes that these values are conservative, and both cats and dogs can benefit from substantially higher protein levels in their food. Proteins are essential for a healthy lifestyle, so the app is quick to adopt these higher values instead of solely restricting to the minima. Research in this area continues at a high pace doi: 10.3390 and advances are integrated into the app as, and when, they are verified and proven to improve feeding quality.
Technologies used in the app are developed in-house by Smarthound. These include a modern CRDT based approach to efficiently synchronise your data. This allows powerful features such as eventual consistency (even when disconnected for long duration), consensus, efficient storage, and minimal (mobile) data requirements. To calculate food blends the app relies on modern artificial intelligence (AI) and machine learning (ML) techniques along with bespoke new technologies for a range of extensions, such as fuzzy and soft constraints.