Nutrition and Health : Accelerate Your Studies with Synthetic Data
In the animal nutrition and health sector, the quality of your data determines the quality of your decisions. Yet between the challenges of building cohorts, missing data, and regulatory constraints, R&D teams often face the same obstacle : data that is too scarce to take their research as far as they would like.
ORIGA is a generative AI platform for clinical data augmentation and imputation, specifically designed to address this challenge.
You lack data. Your studies suffer as a result.
Building a representative animal cohort takes time. A great deal of time. You need to recruit the right species and the right profiles, under the right conditions. And when the data is finally available, it may still be incomplete, imbalanced, or insufficient to train your predictive models with the level of robustness you require.
This is a reality well known to R&D teams in the animal nutrition industry, veterinary laboratories, and animal health companies. And it is not a matter of expertise — it is a structural challenge, inherent to the very nature of these studies.
This scenario, is all too familiar
Imagine this. Your team has been working for several months on modeling puppy growth curves. You want to develop a predictive tool capable of forecasting growth trajectories based on breed, diet, and other key biological parameters.
The data you have collected is real, rigorous, and validated. But it covers only a limited number of breeds, with longitudinal follow-up that is sometimes incomplete. Some measurement points are missing. Some breeds are underrepresented. And no matter how well designed your model is, its performance reaches a ceiling — simply because there are not enough cases for it to learn the full diversity of possible growth trajectories.
Starting a new data collection campaign would take months. Expanding the cohort is costly. And your team is under pressure: deadlines are approaching, and product decisions cannot wait.
This is exactly where ORIGA comes in.
Fewer animals, without compromising scientific rigor
In your sector, the use of real animals remains necessary in one specific context: veterinary clinical trials conducted as part of the evaluation and approval of medicines intended for animals. Outside this strictly regulated framework, every animal involved in a study protocol represents an ethical responsibility, a cost, and often a significant logistical constraint.
The 3Rs principle — Replacement, Reduction, and Refinement — is now both a regulatory requirement in Europe (Directive 2010/63/EU) and an increasingly important expectation among your partners, clients, and investors. Reducing the number of animals involved in your studies without compromising statistical power is no longer just an ideal: it is a direction that leading players in the sector are actively pursuing.
ORIGA enables you to work with smaller real-world cohorts while still obtaining the data richness needed for robust analyses. Synthetic data complements your real data — never replacing it where regulations require otherwise — allowing you to achieve more with less.
Data augmentation : enrich your cohorts without recruiting more animals
ORIGA generates realistic synthetic data from your existing datasets. Using advanced generative AI models — Variational Autoencoders (VAEs) — the platform learns the underlying structure of your biological data and generates new animal profiles that are statistically consistent with your original population.
These synthetic data points are not copies of real individuals. They reproduce the statistical properties of your original data while creating a new, diverse population that can be used immediately.
What this enables you to do :
You can enrich your datasets with profiles from underrepresented breeds. You can simulate growth trajectories that were not sufficiently represented in your original cohort. You can train more robust predictive models on a much larger volume of data — without having to launch a new field data collection campaign.
For animal nutrition, this means more accurate growth models, better-targeted formulations, and more strongly supported R&D decisions.
ORIGA enables you to work with smaller real-world cohorts while still obtaining the data richness needed for robust analyses. Synthetic data complements your real data — never replacing it where regulations require otherwise — allowing you to achieve more with less.
Imputation : no longer letting missing data hold you back
In real-world nutritional studies or animal clinical follow-up, missing data is unavoidable. An animal may drop out of the study. A measurement may not be taken at Day 30. A field partner may send an incomplete file.
Ignorer ces lacunes biaise vos analyses. Les compenser manuellement est chronophage et peu rigoureux. Et supprimer les individus incomplets réduit encore davantage votre cohorte — déjà limitée.
ORIGA offers a different approach: synthetic data imputation. The platform fills in missing data using synthetic values that are consistent with your real-world population. Each imputed value is statistically plausible, mathematically validated, and designed to preserve the integrity of your analyses.
The result: complete, reliable, and analysis-ready datasets — without having to exclude any individual from your cohort.
What this means in practice for your teams
Your data scientists gain access to larger, more balanced datasets for training their models. Your R&D teams can test hypotheses on expanded populations without waiting for new data collection. Your regulatory teams work with traceable, statistically validated data that comply with GDPR and AI Act requirements. And your decision-makers can move faster, supported by a stronger knowledge base.
ORIGA incorporates a rigorous validation process for every generation, including Wasserstein and Kullback–Leibler statistical tests, covariance matrix analysis, privacy testing, and review by a committee of human experts. You do not use synthetic data by default — you use it because it is reliable.
They did it with ORIGA
Royal Canin, a global leader in nutrition for dogs and cats, turned to ORIGA to accelerate the development of its puppy growth models. Faced with incomplete longitudinal data and cohorts that were imbalanced across breeds, the teams needed to enrich their dataset without compromising the rigor of their analyses.
Thanks to ORIGA’s data augmentation and imputation capabilities, the teams were able to generate representative synthetic cohorts, fill in missing data in a statistically consistent way, and accelerate the development of their predictive models.
A real-world use case. A proven technology. A measurable outcome.
Learn more
BOTdesign supports clinical research teams in both human and animal health in the co-design and implementation of hybrid real-world/synthetic study protocols. To discuss the feasibility of your project and receive a tailored assessment: direction@botdesign.net