JADBio is an information technology company based in the Greece and the US, focused on BioMed and Multi-omics. JADBio stands for Just Add Data, and the company aims to make machine learning accessible to all regardless of expertise or programming skills. Its purpose-built AI AutoML platform is said to enable researchers to build and deploy accurate and interpretable predictive models with speed and ease.
In a recent experiment, researchers at JADBio collected data from 478 potato samples from potatoes grown in Germany (including climate, soil, and metabolic profiles) in order to create a model capable of differentiating potatoes that resist bruising from those that don’t, and also to predict the potatoes’ susceptibility to acrylamide formation during chip/crisp processing.
JADBio reports that the data was analyzed in only 30 seconds and produced an executable model with numerous performance metrics, including an AUC (Area Under the Curve) of 80%. Eight out of the original 200+ features that were measured per sample were collectively identified as significant in predicting potato quality. These eight predictors were then assembled into six equivalent biosignatures.
The JADBio platform was applied to Steinfach et al.’s metabolomics profile and climate/soil data, and the company says it produced accurate predictive models that are on par with the ones presented in the Steinfach publication. Find the publication here – titled “Discovering plant metabolic biomarkers for phenotype prediction using an untargeted approach“.
According to JADBio, the results of the experiment shows its AI platform can produce a diagnostic model to predict the quality of potato chips in an automated way.